fix: allow env overrides to create missing config keys (#2064)

Previously, environment variable overrides (e.g. SYSTEM__INSTANCE_ID)
were silently skipped if the target key didn't already exist in
data/config.yaml. This caused SaaS pods running older LangBot images
(whose config template lacked system.instance_id) to ignore the
SYSTEM__INSTANCE_ID env var, falling back to a random UUID that
didn't match the pod UUID — breaking idle timeout tracking.

Now env overrides create missing keys (as strings) and missing
intermediate dicts, so they work regardless of template version.

Co-authored-by: rocksclawbot <rocksclawbot@users.noreply.github.com>

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "langbot"
-version = "4.8.6"
+version = "4.9.3"
 description = "Production-grade platform for building agentic IM bots"
 readme = "README.md"
 license-files = ["LICENSE"]
@@ -61,16 +61,17 @@ dependencies = [
     "html2text>=2024.2.26",
     "langchain>=0.2.0",
     "langchain-text-splitters>=0.0.1",
-    "chromadb>=0.4.24",
+    "chromadb>=1.0.0,<2.0.0",
     "qdrant-client (>=1.15.1,<2.0.0)",
-    "pyseekdb==1.0.0b7",
-    "langbot-plugin==0.2.7",
+    "pyseekdb==1.1.0.post3",
+    "langbot-plugin==0.3.2",
     "asyncpg>=0.30.0",
     "line-bot-sdk>=3.19.0",
     "tboxsdk>=0.0.10",
     "boto3>=1.35.0",
     "pymilvus>=2.6.4",
     "pgvector>=0.4.1",
+    "botocore>=1.42.39",
 ]
 keywords = [
     "bot",

src/langbot/__init__.py

@@ -1,3 +1,3 @@
 """LangBot - Production-grade platform for building agentic IM bots"""
 
-__version__ = '4.8.6'
+__version__ = '4.9.3'

src/langbot/libs/wechatpad_api/util/http_util.py

@@ -1,5 +1,5 @@
 import requests
-import aiohttp
+from langbot.pkg.utils import httpclient
 
 
 def post_json(base_url, token, data=None):
@@ -63,16 +63,16 @@ async def async_request(
     """
     headers = {'Content-Type': 'application/json'}
     url = f'{base_url}?key={token_key}'
-    async with aiohttp.ClientSession() as session:
-        async with session.request(
-            method=method, url=url, params=params, headers=headers, data=data, json=json
-        ) as response:
-            response.raise_for_status()  # 如果状态码不是200，抛出异常
-            result = await response.json()
-            # print(result)
-            return result
-            # if result.get('Code') == 200:
-            #
-            #     return await result
-            # else:
-            #     raise RuntimeError("请求失败",response.text)
+    session = httpclient.get_session()
+    async with session.request(
+        method=method, url=url, params=params, headers=headers, data=data, json=json
+    ) as response:
+        response.raise_for_status()  # 如果状态码不是200，抛出异常
+        result = await response.json()
+        # print(result)
+        return result
+        # if result.get('Code') == 200:
+        #
+        #     return await result
+        # else:
+        #     raise RuntimeError("请求失败",response.text)

src/langbot/libs/wecom_ai_bot_api/api.py

@@ -199,6 +199,253 @@ class StreamSessionManager:
                 self._msg_index.pop(msg_id, None)
 
 
+async def download_encrypted_file(download_url: str, encoding_aes_key: str, logger: EventLogger) -> Optional[str]:
+    """Download an AES-encrypted file from WeChat Work and return as data URI.
+
+    Args:
+        download_url: The encrypted file download URL.
+        encoding_aes_key: The AES key used for decryption (base64-encoded, without trailing '=').
+        logger: Logger instance.
+
+    Returns:
+        A data URI string (e.g. 'data:image/jpeg;base64,...') or None on failure.
+    """
+    if not download_url:
+        return None
+    async with httpx.AsyncClient() as client:
+        response = await client.get(download_url)
+        if response.status_code != 200:
+            await logger.error(f'failed to get file: {response.text}')
+            return None
+        encrypted_bytes = response.content
+
+    aes_key = base64.b64decode(encoding_aes_key + '=')
+    iv = aes_key[:16]
+
+    cipher = AES.new(aes_key, AES.MODE_CBC, iv)
+    decrypted = cipher.decrypt(encrypted_bytes)
+
+    pad_len = decrypted[-1]
+    decrypted = decrypted[:-pad_len]
+
+    if decrypted.startswith(b'\xff\xd8'):
+        mime_type = 'image/jpeg'
+    elif decrypted.startswith(b'\x89PNG'):
+        mime_type = 'image/png'
+    elif decrypted.startswith((b'GIF87a', b'GIF89a')):
+        mime_type = 'image/gif'
+    elif decrypted.startswith(b'BM'):
+        mime_type = 'image/bmp'
+    elif decrypted.startswith(b'II*\x00') or decrypted.startswith(b'MM\x00*'):
+        mime_type = 'image/tiff'
+    else:
+        mime_type = 'application/octet-stream'
+
+    base64_str = base64.b64encode(decrypted).decode('utf-8')
+    return f'data:{mime_type};base64,{base64_str}'
+
+
+async def parse_wecom_bot_message(
+    msg_json: dict[str, Any], encoding_aes_key: str, logger: EventLogger
+) -> dict[str, Any]:
+    """Parse a decrypted WeChat Work AI Bot message JSON into a unified message dict.
+
+    This is the shared message parsing logic used by both webhook and WebSocket modes.
+
+    Args:
+        msg_json: The decrypted message JSON from WeChat Work.
+        encoding_aes_key: AES key for file decryption.
+        logger: Logger instance.
+
+    Returns:
+        A dict suitable for constructing a WecomBotEvent.
+    """
+    message_data: dict[str, Any] = {}
+
+    msg_type = msg_json.get('msgtype', '')
+    if msg_type:
+        message_data['msgtype'] = msg_type
+
+    if msg_json.get('chattype', '') == 'single':
+        message_data['type'] = 'single'
+    elif msg_json.get('chattype', '') == 'group':
+        message_data['type'] = 'group'
+
+    max_inline_file_size = 5 * 1024 * 1024
+
+    async def _safe_download(url: str):
+        if not url:
+            return None
+        return await download_encrypted_file(url, encoding_aes_key, logger)
+
+    if msg_type == 'text':
+        message_data['content'] = msg_json.get('text', {}).get('content')
+    elif msg_type == 'markdown':
+        message_data['content'] = msg_json.get('markdown', {}).get('content') or msg_json.get('text', {}).get(
+            'content', ''
+        )
+    elif msg_type == 'image':
+        picurl = msg_json.get('image', {}).get('url', '')
+        base64_data = await _safe_download(picurl)
+        if base64_data:
+            message_data['picurl'] = base64_data
+            message_data['images'] = [base64_data]
+    elif msg_type == 'voice':
+        voice_info = msg_json.get('voice', {}) or {}
+        download_url = voice_info.get('url')
+        message_data['voice'] = {
+            'url': download_url,
+            'md5sum': voice_info.get('md5sum') or voice_info.get('md5'),
+            'filesize': voice_info.get('filesize') or voice_info.get('size'),
+            'sdkfileid': voice_info.get('sdkfileid') or voice_info.get('fileid'),
+        }
+        if voice_info.get('content'):
+            message_data['content'] = voice_info.get('content')
+        if (message_data['voice'].get('filesize') or 0) <= max_inline_file_size:
+            voice_base64 = await _safe_download(download_url)
+            if voice_base64:
+                message_data['voice']['base64'] = voice_base64
+    elif msg_type == 'video':
+        video_info = msg_json.get('video', {}) or {}
+        download_url = video_info.get('url')
+        video_data = {
+            'url': download_url,
+            'filesize': video_info.get('filesize') or video_info.get('size'),
+            'sdkfileid': video_info.get('sdkfileid') or video_info.get('fileid'),
+            'md5sum': video_info.get('md5sum') or video_info.get('md5'),
+            'filename': video_info.get('filename') or video_info.get('name'),
+        }
+        if (video_data.get('filesize') or 0) <= max_inline_file_size:
+            video_base64 = await _safe_download(download_url)
+            if video_base64:
+                video_data['base64'] = video_base64
+        message_data['video'] = video_data
+    elif msg_type == 'file':
+        file_info = msg_json.get('file', {}) or {}
+        download_url = file_info.get('url') or file_info.get('fileurl')
+        file_data = {
+            'filename': file_info.get('filename') or file_info.get('name'),
+            'filesize': file_info.get('filesize') or file_info.get('size'),
+            'md5sum': file_info.get('md5sum') or file_info.get('md5'),
+            'sdkfileid': file_info.get('sdkfileid') or file_info.get('fileid'),
+            'download_url': download_url,
+            'extra': file_info,
+        }
+        if (file_data.get('filesize') or 0) <= max_inline_file_size:
+            file_base64 = await _safe_download(download_url)
+            if file_base64:
+                file_data['base64'] = file_base64
+        message_data['file'] = file_data
+    elif msg_type == 'link':
+        message_data['link'] = msg_json.get('link', {})
+        if not message_data.get('content'):
+            title = message_data['link'].get('title', '')
+            desc = message_data['link'].get('description') or message_data['link'].get('digest', '')
+            message_data['content'] = '\n'.join(filter(None, [title, desc]))
+    elif msg_type == 'mixed':
+        items = msg_json.get('mixed', {}).get('msg_item', [])
+        texts = []
+        images = []
+        files = []
+        voices = []
+        videos = []
+        links = []
+        for item in items:
+            item_type = item.get('msgtype')
+            if item_type == 'text':
+                texts.append(item.get('text', {}).get('content', ''))
+            elif item_type == 'image':
+                img_url = item.get('image', {}).get('url')
+                base64_data = await _safe_download(img_url)
+                if base64_data:
+                    images.append(base64_data)
+            elif item_type == 'file':
+                file_info = item.get('file', {}) or {}
+                download_url = file_info.get('url') or file_info.get('fileurl')
+                file_data = {
+                    'filename': file_info.get('filename') or file_info.get('name'),
+                    'filesize': file_info.get('filesize') or file_info.get('size'),
+                    'md5sum': file_info.get('md5sum') or file_info.get('md5'),
+                    'sdkfileid': file_info.get('sdkfileid') or file_info.get('fileid'),
+                    'download_url': download_url,
+                    'extra': file_info,
+                }
+                if (file_data.get('filesize') or 0) <= max_inline_file_size:
+                    file_base64 = await _safe_download(download_url)
+                    if file_base64:
+                        file_data['base64'] = file_base64
+                files.append(file_data)
+            elif item_type == 'voice':
+                voice_info = item.get('voice', {}) or {}
+                download_url = voice_info.get('url')
+                voice_data = {
+                    'url': download_url,
+                    'md5sum': voice_info.get('md5sum') or voice_info.get('md5'),
+                    'filesize': voice_info.get('filesize') or voice_info.get('size'),
+                    'sdkfileid': voice_info.get('sdkfileid') or voice_info.get('fileid'),
+                }
+                if voice_info.get('content'):
+                    texts.append(voice_info.get('content'))
+                if (voice_data.get('filesize') or 0) <= max_inline_file_size:
+                    voice_base64 = await _safe_download(download_url)
+                    if voice_base64:
+                        voice_data['base64'] = voice_base64
+                voices.append(voice_data)
+            elif item_type == 'video':
+                video_info = item.get('video', {}) or {}
+                download_url = video_info.get('url')
+                video_data = {
+                    'url': download_url,
+                    'filesize': video_info.get('filesize') or video_info.get('size'),
+                    'sdkfileid': video_info.get('sdkfileid') or video_info.get('fileid'),
+                    'md5sum': video_info.get('md5sum') or video_info.get('md5'),
+                    'filename': video_info.get('filename') or video_info.get('name'),
+                }
+                if (video_data.get('filesize') or 0) <= max_inline_file_size:
+                    video_base64 = await _safe_download(download_url)
+                    if video_base64:
+                        video_data['base64'] = video_base64
+                videos.append(video_data)
+            elif item_type == 'link':
+                links.append(item.get('link', {}))
+
+        if texts:
+            message_data['content'] = ' '.join(texts)
+        if images:
+            message_data['images'] = images
+            message_data['picurl'] = images[0]
+        if files:
+            message_data['files'] = files
+            message_data['file'] = files[0]
+        if voices:
+            message_data['voices'] = voices
+            message_data['voice'] = voices[0]
+        if videos:
+            message_data['videos'] = videos
+            message_data['video'] = videos[0]
+        if links:
+            message_data['link'] = links[0]
+        if items:
+            message_data['attachments'] = items
+    else:
+        message_data['raw_msg'] = msg_json
+
+    from_info = msg_json.get('from', {})
+    message_data['userid'] = from_info.get('userid', '')
+    message_data['username'] = from_info.get('alias', '') or from_info.get('name', '') or from_info.get('userid', '')
+
+    if msg_json.get('chattype', '') == 'group':
+        message_data['chatid'] = msg_json.get('chatid', '')
+        message_data['chatname'] = msg_json.get('chatname', '') or msg_json.get('chatid', '')
+
+    message_data['msgid'] = msg_json.get('msgid', '')
+
+    if msg_json.get('aibotid'):
+        message_data['aibotid'] = msg_json.get('aibotid', '')
+
+    return message_data
+
+
 class WecomBotClient:
     def __init__(self, Token: str, EnCodingAESKey: str, Corpid: str, logger: EventLogger, unified_mode: bool = False):
         """企业微信智能机器人客户端。
@@ -455,196 +702,7 @@ class WecomBotClient:
         return await self._handle_post_initial_response(msg_json, nonce)
 
     async def get_message(self, msg_json):
-        message_data = {}
-
-        msg_type = msg_json.get('msgtype', '')
-        if msg_type:
-            message_data['msgtype'] = msg_type
-
-        if msg_json.get('chattype', '') == 'single':
-            message_data['type'] = 'single'
-        elif msg_json.get('chattype', '') == 'group':
-            message_data['type'] = 'group'
-
-        max_inline_file_size = 5 * 1024 * 1024  # avoid decoding very large payloads by default
-
-        async def _safe_download(url: str):
-            if not url:
-                return None
-            return await self.download_url_to_base64(url, self.EnCodingAESKey)
-
-        if msg_type == 'text':
-            message_data['content'] = msg_json.get('text', {}).get('content')
-        elif msg_type == 'markdown':
-            message_data['content'] = msg_json.get('markdown', {}).get('content') or msg_json.get('text', {}).get(
-                'content', ''
-            )
-        elif msg_type == 'image':
-            picurl = msg_json.get('image', {}).get('url', '')
-            base64_data = await _safe_download(picurl)
-            if base64_data:
-                message_data['picurl'] = base64_data
-                message_data['images'] = [base64_data]
-        elif msg_type == 'voice':
-            voice_info = msg_json.get('voice', {}) or {}
-            download_url = voice_info.get('url')
-            message_data['voice'] = {
-                'url': download_url,
-                'md5sum': voice_info.get('md5sum') or voice_info.get('md5'),
-                'filesize': voice_info.get('filesize') or voice_info.get('size'),
-                'sdkfileid': voice_info.get('sdkfileid') or voice_info.get('fileid'),
-            }
-            # 企业微信智能转写文本（如果已有）直接复用，避免重复转写
-            if voice_info.get('content'):
-                message_data['content'] = voice_info.get('content')
-            if (message_data['voice'].get('filesize') or 0) <= max_inline_file_size:
-                voice_base64 = await _safe_download(download_url)
-                if voice_base64:
-                    message_data['voice']['base64'] = voice_base64
-        elif msg_type == 'video':
-            video_info = msg_json.get('video', {}) or {}
-            download_url = video_info.get('url')
-            video_data = {
-                'url': download_url,
-                'filesize': video_info.get('filesize') or video_info.get('size'),
-                'sdkfileid': video_info.get('sdkfileid') or video_info.get('fileid'),
-                'md5sum': video_info.get('md5sum') or video_info.get('md5'),
-                'filename': video_info.get('filename') or video_info.get('name'),
-            }
-            if (video_data.get('filesize') or 0) <= max_inline_file_size:
-                video_base64 = await _safe_download(download_url)
-                if video_base64:
-                    video_data['base64'] = video_base64
-            message_data['video'] = video_data
-        elif msg_type == 'file':
-            file_info = msg_json.get('file', {}) or {}
-            download_url = file_info.get('url') or file_info.get('fileurl')
-            file_data = {
-                'filename': file_info.get('filename') or file_info.get('name'),
-                'filesize': file_info.get('filesize') or file_info.get('size'),
-                'md5sum': file_info.get('md5sum') or file_info.get('md5'),
-                'sdkfileid': file_info.get('sdkfileid') or file_info.get('fileid'),
-                'download_url': download_url,
-                'extra': file_info,
-            }
-            if (file_data.get('filesize') or 0) <= max_inline_file_size:
-                file_base64 = await _safe_download(download_url)
-                if file_base64:
-                    file_data['base64'] = file_base64
-            message_data['file'] = file_data
-        elif msg_type == 'link':
-            message_data['link'] = msg_json.get('link', {})
-            if not message_data.get('content'):
-                title = message_data['link'].get('title', '')
-                desc = message_data['link'].get('description') or message_data['link'].get('digest', '')
-                message_data['content'] = '\n'.join(filter(None, [title, desc]))
-        elif msg_type == 'mixed':
-            items = msg_json.get('mixed', {}).get('msg_item', [])
-            texts = []
-            images = []
-            files = []
-            voices = []
-            videos = []
-            links = []
-            for item in items:
-                item_type = item.get('msgtype')
-                if item_type == 'text':
-                    texts.append(item.get('text', {}).get('content', ''))
-                elif item_type == 'image':
-                    img_url = item.get('image', {}).get('url')
-                    base64_data = await _safe_download(img_url)
-                    if base64_data:
-                        images.append(base64_data)
-                elif item_type == 'file':
-                    file_info = item.get('file', {}) or {}
-                    download_url = file_info.get('url') or file_info.get('fileurl')
-                    file_data = {
-                        'filename': file_info.get('filename') or file_info.get('name'),
-                        'filesize': file_info.get('filesize') or file_info.get('size'),
-                        'md5sum': file_info.get('md5sum') or file_info.get('md5'),
-                        'sdkfileid': file_info.get('sdkfileid') or file_info.get('fileid'),
-                        'download_url': download_url,
-                        'extra': file_info,
-                    }
-                    if (file_data.get('filesize') or 0) <= max_inline_file_size:
-                        file_base64 = await _safe_download(download_url)
-                        if file_base64:
-                            file_data['base64'] = file_base64
-                    files.append(file_data)
-                elif item_type == 'voice':
-                    voice_info = item.get('voice', {}) or {}
-                    download_url = voice_info.get('url')
-                    voice_data = {
-                        'url': download_url,
-                        'md5sum': voice_info.get('md5sum') or voice_info.get('md5'),
-                        'filesize': voice_info.get('filesize') or voice_info.get('size'),
-                        'sdkfileid': voice_info.get('sdkfileid') or voice_info.get('fileid'),
-                    }
-                    if voice_info.get('content'):
-                        texts.append(voice_info.get('content'))
-                    if (voice_data.get('filesize') or 0) <= max_inline_file_size:
-                        voice_base64 = await _safe_download(download_url)
-                        if voice_base64:
-                            voice_data['base64'] = voice_base64
-                    voices.append(voice_data)
-                elif item_type == 'video':
-                    video_info = item.get('video', {}) or {}
-                    download_url = video_info.get('url')
-                    video_data = {
-                        'url': download_url,
-                        'filesize': video_info.get('filesize') or video_info.get('size'),
-                        'sdkfileid': video_info.get('sdkfileid') or video_info.get('fileid'),
-                        'md5sum': video_info.get('md5sum') or video_info.get('md5'),
-                        'filename': video_info.get('filename') or video_info.get('name'),
-                    }
-                    if (video_data.get('filesize') or 0) <= max_inline_file_size:
-                        video_base64 = await _safe_download(download_url)
-                        if video_base64:
-                            video_data['base64'] = video_base64
-                    videos.append(video_data)
-                elif item_type == 'link':
-                    links.append(item.get('link', {}))
-
-            if texts:
-                message_data['content'] = ' '.join(texts)  # 拼接所有 text
-            if images:
-                message_data['images'] = images
-                message_data['picurl'] = images[0]  # 只保留第一个 image
-            if files:
-                message_data['files'] = files
-                message_data['file'] = files[0]
-            if voices:
-                message_data['voices'] = voices
-                message_data['voice'] = voices[0]
-            if videos:
-                message_data['videos'] = videos
-                message_data['video'] = videos[0]
-            if links:
-                message_data['link'] = links[0]
-            if items:
-                message_data['attachments'] = items
-        else:
-            message_data['raw_msg'] = msg_json
-
-        # Extract user information
-        from_info = msg_json.get('from', {})
-        message_data['userid'] = from_info.get('userid', '')
-        message_data['username'] = (
-            from_info.get('alias', '') or from_info.get('name', '') or from_info.get('userid', '')
-        )
-
-        # Extract chat/group information
-        if msg_json.get('chattype', '') == 'group':
-            message_data['chatid'] = msg_json.get('chatid', '')
-            # Try to get group name if available
-            message_data['chatname'] = msg_json.get('chatname', '') or msg_json.get('chatid', '')
-
-        message_data['msgid'] = msg_json.get('msgid', '')
-
-        if msg_json.get('aibotid'):
-            message_data['aibotid'] = msg_json.get('aibotid', '')
-
-        return message_data
+        return await parse_wecom_bot_message(msg_json, self.EnCodingAESKey, self.logger)
 
     async def _handle_message(self, event: wecombotevent.WecomBotEvent):
         """
@@ -712,39 +770,7 @@ class WecomBotClient:
         return decorator
 
     async def download_url_to_base64(self, download_url, encoding_aes_key):
-        async with httpx.AsyncClient() as client:
-            response = await client.get(download_url)
-            if response.status_code != 200:
-                await self.logger.error(f'failed to get file: {response.text}')
-                return None
-
-            encrypted_bytes = response.content
-
-        aes_key = base64.b64decode(encoding_aes_key + '=')  # base64 补齐
-        iv = aes_key[:16]
-
-        cipher = AES.new(aes_key, AES.MODE_CBC, iv)
-        decrypted = cipher.decrypt(encrypted_bytes)
-
-        pad_len = decrypted[-1]
-        decrypted = decrypted[:-pad_len]
-
-        if decrypted.startswith(b'\xff\xd8'):  # JPEG
-            mime_type = 'image/jpeg'
-        elif decrypted.startswith(b'\x89PNG'):  # PNG
-            mime_type = 'image/png'
-        elif decrypted.startswith((b'GIF87a', b'GIF89a')):  # GIF
-            mime_type = 'image/gif'
-        elif decrypted.startswith(b'BM'):  # BMP
-            mime_type = 'image/bmp'
-        elif decrypted.startswith(b'II*\x00') or decrypted.startswith(b'MM\x00*'):  # TIFF
-            mime_type = 'image/tiff'
-        else:
-            mime_type = 'application/octet-stream'
-
-        # 转 base64
-        base64_str = base64.b64encode(decrypted).decode('utf-8')
-        return f'data:{mime_type};base64,{base64_str}'
+        return await download_encrypted_file(download_url, encoding_aes_key, self.logger)
 
     async def run_task(self, host: str, port: int, *args, **kwargs):
         """

src/langbot/libs/wecom_ai_bot_api/ws_client.py

@@ -0,0 +1,596 @@
+"""WeChat Work AI Bot WebSocket long connection client.
+
+Implements the WebSocket protocol for receiving messages and sending replies
+via a persistent connection to wss://openws.work.weixin.qq.com, as an
+alternative to the HTTP callback (webhook) mode.
+
+Protocol reference: https://developer.work.weixin.qq.com/document/path/101463
+Official Node.js SDK: https://github.com/WecomTeam/aibot-node-sdk
+"""
+
+from __future__ import annotations
+
+import asyncio
+import json
+import secrets
+import time
+import traceback
+from typing import Any, Callable, Optional
+
+import aiohttp
+
+from langbot.libs.wecom_ai_bot_api import wecombotevent
+from langbot.libs.wecom_ai_bot_api.api import parse_wecom_bot_message
+from langbot.pkg.platform.logger import EventLogger
+
+DEFAULT_WS_URL = 'wss://openws.work.weixin.qq.com'
+
+# WebSocket frame command constants
+CMD_SUBSCRIBE = 'aibot_subscribe'
+CMD_HEARTBEAT = 'ping'
+CMD_MSG_CALLBACK = 'aibot_msg_callback'
+CMD_EVENT_CALLBACK = 'aibot_event_callback'
+CMD_RESPOND_MSG = 'aibot_respond_msg'
+CMD_RESPOND_WELCOME = 'aibot_respond_welcome_msg'
+CMD_RESPOND_UPDATE = 'aibot_respond_update_msg'
+CMD_SEND_MSG = 'aibot_send_msg'
+
+
+def _generate_req_id(prefix: str) -> str:
+    """Generate a unique request ID in the format: {prefix}_{timestamp}_{random}."""
+    ts = int(time.time() * 1000)
+    rand = secrets.token_hex(4)
+    return f'{prefix}_{ts}_{rand}'
+
+
+class WecomBotWsClient:
+    """WeChat Work AI Bot WebSocket long connection client.
+
+    Provides message receiving, streaming reply, proactive message sending,
+    and event callback handling over a persistent WebSocket connection.
+    """
+
+    def __init__(
+        self,
+        bot_id: str,
+        secret: str,
+        logger: EventLogger,
+        encoding_aes_key: str = '',
+        ws_url: str = DEFAULT_WS_URL,
+        heartbeat_interval: float = 30.0,
+        max_reconnect_attempts: int = -1,
+        reconnect_base_delay: float = 1.0,
+        reconnect_max_delay: float = 30.0,
+    ):
+        self.bot_id = bot_id
+        self.secret = secret
+        self.logger = logger
+        self.encoding_aes_key = encoding_aes_key
+        self.ws_url = ws_url
+        self.heartbeat_interval = heartbeat_interval
+        self.max_reconnect_attempts = max_reconnect_attempts
+        self.reconnect_base_delay = reconnect_base_delay
+        self.reconnect_max_delay = reconnect_max_delay
+
+        self._ws: Optional[aiohttp.ClientWebSocketResponse] = None
+        self._session: Optional[aiohttp.ClientSession] = None
+        self._running = False
+        self._heartbeat_task: Optional[asyncio.Task] = None
+        self._missed_pong_count = 0
+        self._max_missed_pong = 2
+        self._reconnect_attempts = 0
+
+        # Message handler registry (same pattern as WecomBotClient)
+        self._message_handlers: dict[str, list[Callable]] = {}
+        # Message deduplication
+        self._msg_id_map: dict[str, int] = {}
+
+        # Pending ACK futures: req_id -> Future[dict]
+        self._pending_acks: dict[str, asyncio.Future] = {}
+        # Per-req_id serial reply queues
+        self._reply_queues: dict[str, asyncio.Queue] = {}
+        self._reply_workers: dict[str, asyncio.Task] = {}
+        self._reply_ack_timeout = 5.0
+
+        # Stream ID tracking for WebSocket mode
+        self._stream_ids: dict[str, str] = {}  # msg_id -> req_id|stream_id
+        # Dedup: skip sending when content hasn't changed
+        self._stream_last_content: dict[str, str] = {}  # msg_id -> last content sent
+
+    # ── Public API ──────────────────────────────────────────────────
+
+    async def connect(self):
+        """Connect to WebSocket server with automatic reconnection.
+
+        This method blocks until disconnect() is called or max reconnect
+        attempts are exhausted.
+        """
+        self._running = True
+        self._reconnect_attempts = 0
+
+        while self._running:
+            try:
+                await self._connect_once()
+            except Exception:
+                if not self._running:
+                    break
+                await self.logger.error(f'WebSocket connection error: {traceback.format_exc()}')
+
+            if not self._running:
+                break
+
+            # Reconnect with exponential backoff
+            if self.max_reconnect_attempts != -1 and self._reconnect_attempts >= self.max_reconnect_attempts:
+                await self.logger.error(f'Max reconnect attempts reached ({self.max_reconnect_attempts}), giving up')
+                break
+
+            self._reconnect_attempts += 1
+            delay = min(
+                self.reconnect_base_delay * (2 ** (self._reconnect_attempts - 1)),
+                self.reconnect_max_delay,
+            )
+            await self.logger.info(f'Reconnecting in {delay:.1f}s (attempt {self._reconnect_attempts})...')
+            await asyncio.sleep(delay)
+
+    async def disconnect(self):
+        """Gracefully disconnect from the WebSocket server."""
+        self._running = False
+        if self._heartbeat_task and not self._heartbeat_task.done():
+            self._heartbeat_task.cancel()
+        for task in self._reply_workers.values():
+            if not task.done():
+                task.cancel()
+        if self._ws and not self._ws.closed:
+            await self._ws.close()
+        self._ws = None
+        if self._session and not self._session.closed:
+            await self._session.close()
+        self._session = None
+
+    def on_message(self, msg_type: str) -> Callable:
+        """Decorator to register a message handler.
+
+        Same interface as WecomBotClient.on_message for compatibility.
+
+        Args:
+            msg_type: 'single', 'group', or specific message type.
+        """
+
+        def decorator(func: Callable[[wecombotevent.WecomBotEvent], Any]):
+            if msg_type not in self._message_handlers:
+                self._message_handlers[msg_type] = []
+            self._message_handlers[msg_type].append(func)
+            return func
+
+        return decorator
+
+    async def reply_stream(
+        self,
+        req_id: str,
+        stream_id: str,
+        content: str,
+        finish: bool = False,
+    ) -> Optional[dict]:
+        """Send a streaming reply frame.
+
+        Args:
+            req_id: The req_id from the original message frame (must be passed through).
+            stream_id: The stream ID for this streaming session.
+            content: The content to send (supports Markdown).
+            finish: Whether this is the final chunk.
+
+        Returns:
+            The ACK frame dict, or None on failure.
+        """
+        body = {
+            'msgtype': 'stream',
+            'stream': {
+                'id': stream_id,
+                'finish': finish,
+                'content': content,
+            },
+        }
+        return await self._send_reply(req_id, body)
+
+    async def reply_text(self, req_id: str, content: str) -> Optional[dict]:
+        """Send a non-streaming text reply.
+
+        Args:
+            req_id: The req_id from the original message frame.
+            content: The text content to reply.
+
+        Returns:
+            The ACK frame dict, or None on failure.
+        """
+        body = {
+            'msgtype': 'markdown',
+            'markdown': {
+                'content': content,
+            },
+        }
+        return await self._send_reply(req_id, body)
+
+    async def send_message(self, chat_id: str, content: str, msgtype: str = 'markdown') -> Optional[dict]:
+        """Proactively send a message to a specified chat.
+
+        Args:
+            chat_id: The chat ID (userid for single chat, chatid for group chat).
+            content: The message content.
+            msgtype: Message type, 'markdown' by default.
+
+        Returns:
+            The ACK frame dict, or None on failure.
+        """
+        req_id = _generate_req_id(CMD_SEND_MSG)
+        body: dict[str, Any] = {
+            'chatid': chat_id,
+            'msgtype': msgtype,
+        }
+        if msgtype == 'markdown':
+            body['markdown'] = {'content': content}
+        elif msgtype == 'text':
+            body['text'] = {'content': content}
+        return await self._send_reply(req_id, body, cmd=CMD_SEND_MSG)
+
+    async def push_stream_chunk(self, msg_id: str, content: str, is_final: bool = False) -> bool:
+        """Push a streaming chunk for a given message ID.
+
+        Compatible interface with WecomBotClient.push_stream_chunk.
+
+        Args:
+            msg_id: The original message ID.
+            content: The cumulative content from the pipeline.
+            is_final: Whether this is the final chunk.
+
+        Returns:
+            True if the stream session exists and chunk was sent.
+        """
+        key = self._stream_ids.get(msg_id)
+        if not key:
+            return False
+        req_id, stream_id = key.split('|', 1)
+        try:
+            # Skip sending if content hasn't changed (e.g. during tool call argument streaming)
+            if not is_final and content == self._stream_last_content.get(msg_id):
+                return True
+            await self.reply_stream(req_id, stream_id, content, finish=is_final)
+            self._stream_last_content[msg_id] = content
+            if is_final:
+                self._stream_ids.pop(msg_id, None)
+                self._stream_last_content.pop(msg_id, None)
+            return True
+        except Exception:
+            await self.logger.error(f'Failed to push stream chunk: {traceback.format_exc()}')
+            return False
+
+    async def set_message(self, msg_id: str, content: str):
+        """Fallback: send content as a final stream chunk or direct reply.
+
+        Compatible interface with WecomBotClient.set_message.
+        """
+        handled = await self.push_stream_chunk(msg_id, content, is_final=True)
+        if not handled:
+            await self.logger.warning(f'No active stream for msg_id={msg_id}, message dropped')
+
+    # ── Connection lifecycle ────────────────────────────────────────
+
+    async def _connect_once(self):
+        """Establish a single WebSocket connection, authenticate, and listen."""
+        await self.logger.info(f'Connecting to {self.ws_url}...')
+
+        self._session = aiohttp.ClientSession()
+        try:
+            self._ws = await self._session.ws_connect(self.ws_url)
+            self._missed_pong_count = 0
+            self._reconnect_attempts = 0
+            await self.logger.info('WebSocket connected, sending auth...')
+
+            await self._send_auth()
+
+            # Wait for auth response
+            auth_ok = await self._wait_for_auth()
+            if not auth_ok:
+                await self.logger.error('Authentication failed')
+                return
+
+            await self.logger.info('Authenticated successfully')
+
+            # Start heartbeat
+            self._heartbeat_task = asyncio.create_task(self._heartbeat_loop())
+
+            try:
+                await self._listen_loop()
+            finally:
+                if self._heartbeat_task and not self._heartbeat_task.done():
+                    self._heartbeat_task.cancel()
+                self._clear_pending_acks('Connection closed')
+        finally:
+            if self._ws and not self._ws.closed:
+                await self._ws.close()
+            self._ws = None
+            if self._session and not self._session.closed:
+                await self._session.close()
+            self._session = None
+
+    async def _send_auth(self):
+        """Send the authentication frame."""
+        frame = {
+            'cmd': CMD_SUBSCRIBE,
+            'headers': {'req_id': _generate_req_id(CMD_SUBSCRIBE)},
+            'body': {
+                'bot_id': self.bot_id,
+                'secret': self.secret,
+            },
+        }
+        await self._send_frame(frame)
+
+    async def _wait_for_auth(self) -> bool:
+        """Wait for and validate the authentication response."""
+        try:
+            msg = await asyncio.wait_for(self._ws.receive(), timeout=10.0)
+            if msg.type in (aiohttp.WSMsgType.TEXT,):
+                frame = json.loads(msg.data)
+                req_id = frame.get('headers', {}).get('req_id', '')
+                if req_id.startswith(CMD_SUBSCRIBE) and frame.get('errcode') == 0:
+                    return True
+                await self.logger.error(f'Auth response: errcode={frame.get("errcode")}, errmsg={frame.get("errmsg")}')
+                return False
+            elif msg.type in (aiohttp.WSMsgType.ERROR, aiohttp.WSMsgType.CLOSED, aiohttp.WSMsgType.CLOSING):
+                await self.logger.error(f'WebSocket closed during auth: {msg.type}')
+                return False
+            await self.logger.error(f'Unexpected message type during auth: {msg.type}')
+            return False
+        except asyncio.TimeoutError:
+            await self.logger.error('Auth response timeout')
+            return False
+
+    async def _heartbeat_loop(self):
+        """Periodically send heartbeat pings."""
+        try:
+            while self._running and self._ws and not self._ws.closed:
+                await asyncio.sleep(self.heartbeat_interval)
+                if not self._running or not self._ws or self._ws.closed:
+                    break
+
+                if self._missed_pong_count >= self._max_missed_pong:
+                    await self.logger.warning(
+                        f'No heartbeat ack for {self._missed_pong_count} consecutive pings, connection considered dead'
+                    )
+                    await self._ws.close()
+                    break
+
+                self._missed_pong_count += 1
+                frame = {
+                    'cmd': CMD_HEARTBEAT,
+                    'headers': {'req_id': _generate_req_id(CMD_HEARTBEAT)},
+                }
+                try:
+                    await self._send_frame(frame)
+                except Exception:
+                    break
+        except asyncio.CancelledError:
+            pass
+
+    async def _listen_loop(self):
+        """Listen for incoming WebSocket frames and dispatch them."""
+        async for msg in self._ws:
+            if not self._running:
+                break
+            if msg.type == aiohttp.WSMsgType.TEXT:
+                try:
+                    frame = json.loads(msg.data)
+                    await self._handle_frame(frame)
+                except json.JSONDecodeError:
+                    await self.logger.error(f'Failed to parse WebSocket message: {str(msg.data)[:200]}')
+                except Exception:
+                    await self.logger.error(f'Error handling frame: {traceback.format_exc()}')
+            elif msg.type == aiohttp.WSMsgType.BINARY:
+                try:
+                    frame = json.loads(msg.data)
+                    await self._handle_frame(frame)
+                except Exception:
+                    await self.logger.error(f'Error handling binary frame: {traceback.format_exc()}')
+            elif msg.type in (aiohttp.WSMsgType.ERROR, aiohttp.WSMsgType.CLOSED, aiohttp.WSMsgType.CLOSING):
+                await self.logger.warning(f'WebSocket connection closed: {msg.type}')
+                break
+
+    # ── Frame handling ──────────────────────────────────────────────
+
+    async def _handle_frame(self, frame: dict):
+        """Route an incoming frame to the appropriate handler."""
+        cmd = frame.get('cmd', '')
+
+        # Message push
+        if cmd == CMD_MSG_CALLBACK:
+            asyncio.create_task(self._handle_message_callback(frame))
+            return
+
+        # Event push
+        if cmd == CMD_EVENT_CALLBACK:
+            asyncio.create_task(self._handle_event_callback(frame))
+            return
+
+        # No cmd → response/ACK frame, dispatch by req_id prefix
+        req_id = frame.get('headers', {}).get('req_id', '')
+
+        # Check pending ACKs first
+        if req_id in self._pending_acks:
+            future = self._pending_acks.pop(req_id)
+            if not future.done():
+                future.set_result(frame)
+            return
+
+        # Heartbeat response
+        if req_id.startswith(CMD_HEARTBEAT):
+            if frame.get('errcode') == 0:
+                self._missed_pong_count = 0
+            return
+
+        # Unknown frame
+        await self.logger.warning(f'Unknown frame: {json.dumps(frame, ensure_ascii=False)[:200]}')
+
+    async def _handle_message_callback(self, frame: dict):
+        """Handle an incoming message callback frame."""
+        try:
+            body = frame.get('body', {})
+            req_id = frame.get('headers', {}).get('req_id', '')
+
+            # Parse message using shared logic
+            message_data = await parse_wecom_bot_message(body, self.encoding_aes_key, self.logger)
+            if not message_data:
+                return
+
+            # Generate stream_id for this message and store the mapping
+            stream_id = _generate_req_id('stream')
+            msg_id = message_data.get('msgid', '')
+            if msg_id:
+                self._stream_ids[msg_id] = f'{req_id}|{stream_id}'
+            message_data['stream_id'] = stream_id
+            message_data['req_id'] = req_id
+
+            event = wecombotevent.WecomBotEvent(message_data)
+            await self._dispatch_event(event)
+        except Exception:
+            await self.logger.error(f'Error in message callback: {traceback.format_exc()}')
+
+    async def _handle_event_callback(self, frame: dict):
+        """Handle an incoming event callback frame (enter_chat, template_card_event, etc.)."""
+        try:
+            body = frame.get('body', {})
+            req_id = frame.get('headers', {}).get('req_id', '')
+
+            event_info = body.get('event', {})
+            event_type = event_info.get('eventtype', '')
+
+            message_data = {
+                'msgtype': 'event',
+                'type': body.get('chattype', 'single'),
+                'event': event_info,
+                'eventtype': event_type,
+                'msgid': body.get('msgid', ''),
+                'aibotid': body.get('aibotid', ''),
+                'req_id': req_id,
+            }
+
+            from_info = body.get('from', {})
+            message_data['userid'] = from_info.get('userid', '')
+            message_data['username'] = from_info.get('alias', '') or from_info.get('userid', '')
+
+            if body.get('chatid'):
+                message_data['chatid'] = body.get('chatid', '')
+
+            event = wecombotevent.WecomBotEvent(message_data)
+
+            # Dispatch to event-specific handlers
+            if event_type in self._message_handlers:
+                for handler in self._message_handlers[event_type]:
+                    await handler(event)
+
+            # Also dispatch to generic 'event' handlers
+            if 'event' in self._message_handlers:
+                for handler in self._message_handlers['event']:
+                    await handler(event)
+
+        except Exception:
+            await self.logger.error(f'Error in event callback: {traceback.format_exc()}')
+
+    async def _dispatch_event(self, event: wecombotevent.WecomBotEvent):
+        """Dispatch a message event to registered handlers with deduplication."""
+        try:
+            message_id = event.message_id
+            if message_id in self._msg_id_map:
+                self._msg_id_map[message_id] += 1
+                return
+            self._msg_id_map[message_id] = 1
+
+            msg_type = event.type
+            if msg_type in self._message_handlers:
+                for handler in self._message_handlers[msg_type]:
+                    await handler(event)
+        except Exception:
+            await self.logger.error(f'Error dispatching event: {traceback.format_exc()}')
+
+    # ── Reply sending with serial queue ─────────────────────────────
+
+    async def _send_reply(
+        self,
+        req_id: str,
+        body: dict,
+        cmd: str = CMD_RESPOND_MSG,
+    ) -> Optional[dict]:
+        """Send a reply frame and wait for ACK.
+
+        Replies with the same req_id are serialized to maintain ordering.
+        """
+        if not self._ws or self._ws.closed:
+            return None
+
+        frame = {
+            'cmd': cmd,
+            'headers': {'req_id': req_id},
+            'body': body,
+        }
+
+        # Ensure serial delivery per req_id
+        if req_id not in self._reply_queues:
+            self._reply_queues[req_id] = asyncio.Queue()
+            self._reply_workers[req_id] = asyncio.create_task(self._reply_queue_worker(req_id))
+
+        future: asyncio.Future = asyncio.get_event_loop().create_future()
+        await self._reply_queues[req_id].put((frame, future))
+        return await future
+
+    async def _reply_queue_worker(self, req_id: str):
+        """Process reply queue items serially for a given req_id."""
+        queue = self._reply_queues[req_id]
+        try:
+            while self._running:
+                try:
+                    frame, future = await asyncio.wait_for(queue.get(), timeout=60.0)
+                except asyncio.TimeoutError:
+                    # Queue idle, clean up worker
+                    break
+
+                try:
+                    ack = await self._send_and_wait_ack(frame)
+                    if not future.done():
+                        future.set_result(ack)
+                except Exception as e:
+                    if not future.done():
+                        future.set_exception(e)
+        except asyncio.CancelledError:
+            pass
+        finally:
+            self._reply_queues.pop(req_id, None)
+            self._reply_workers.pop(req_id, None)
+
+    async def _send_and_wait_ack(self, frame: dict) -> Optional[dict]:
+        """Send a frame and wait for the corresponding ACK."""
+        req_id = frame['headers']['req_id']
+        ack_future: asyncio.Future = asyncio.get_event_loop().create_future()
+        self._pending_acks[req_id] = ack_future
+
+        try:
+            await self._send_frame(frame)
+            result = await asyncio.wait_for(ack_future, timeout=self._reply_ack_timeout)
+            if result.get('errcode', 0) != 0:
+                await self.logger.warning(
+                    f'Reply ACK error: errcode={result.get("errcode")}, errmsg={result.get("errmsg")}'
+                )
+            return result
+        except asyncio.TimeoutError:
+            self._pending_acks.pop(req_id, None)
+            await self.logger.warning(f'Reply ACK timeout ({self._reply_ack_timeout}s) for req_id={req_id}')
+            return None
+
+    async def _send_frame(self, frame: dict):
+        """Send a JSON frame over the WebSocket connection."""
+        if self._ws and not self._ws.closed:
+            await self._ws.send_str(json.dumps(frame, ensure_ascii=False))
+
+    def _clear_pending_acks(self, reason: str):
+        """Reject all pending ACK futures on disconnection."""
+        for req_id, future in self._pending_acks.items():
+            if not future.done():
+                future.set_exception(ConnectionError(reason))
+        self._pending_acks.clear()

src/langbot/libs/wecom_customer_service_api/api.py

@@ -10,6 +10,7 @@ from typing import Callable
 from .wecomcsevent import WecomCSEvent
 import langbot_plugin.api.entities.builtin.platform.message as platform_message
 import aiofiles
+import time
 
 
 class WecomCSClient:
@@ -34,6 +35,10 @@ class WecomCSClient:
         self.unified_mode = unified_mode
         self.app = Quart(__name__)
 
+        # Customer info cache: {external_userid: (info_dict, timestamp)}
+        self._customer_cache: dict[str, tuple[dict, float]] = {}
+        self._cache_ttl = 60  # Cache TTL in seconds (1 minute)
+
         # 只有在非统一模式下才注册独立路由
         if not self.unified_mode:
             self.app.add_url_rule(
@@ -378,3 +383,53 @@ class WecomCSClient:
     async def get_media_id(self, image: platform_message.Image):
         media_id = await self.upload_to_work(image=image)
         return media_id
+
+    async def get_customer_info(self, external_userid: str) -> dict | None:
+        """
+        Get customer information by external_userid with caching.
+
+        Uses a 1-minute cache to avoid repeated API calls for the same user.
+
+        Args:
+            external_userid: The external user ID of the customer.
+
+        Returns:
+            Customer info dict with 'nickname', 'avatar', etc., or None if not found.
+        """
+        # Check cache first
+        current_time = time.time()
+        if external_userid in self._customer_cache:
+            cached_info, cached_time = self._customer_cache[external_userid]
+            if current_time - cached_time < self._cache_ttl:
+                return cached_info
+
+        # Cache miss or expired, fetch from API
+        if not await self.check_access_token():
+            self.access_token = await self.get_access_token(self.secret)
+
+        url = f'{self.base_url}/kf/customer/batchget?access_token={self.access_token}'
+
+        payload = {
+            'external_userid_list': [external_userid],
+        }
+
+        async with httpx.AsyncClient() as client:
+            response = await client.post(url, json=payload)
+            data = response.json()
+
+            if data.get('errcode') in [40014, 42001]:
+                self.access_token = await self.get_access_token(self.secret)
+                return await self.get_customer_info(external_userid)
+
+            if data.get('errcode', 0) != 0:
+                if self.logger:
+                    await self.logger.warning(f'Failed to get customer info: {data}')
+                return None
+
+            customer_list = data.get('customer_list', [])
+            if customer_list:
+                customer_info = customer_list[0]
+                # Store in cache
+                self._customer_cache[external_userid] = (customer_info, current_time)
+                return customer_info
+            return None

src/langbot/pkg/api/http/controller/groups/knowledge/base.py

@@ -13,7 +13,10 @@ class KnowledgeBaseRouterGroup(group.RouterGroup):
 
             elif quart.request.method == 'POST':
                 json_data = await quart.request.json
-                knowledge_base_uuid = await self.ap.knowledge_service.create_knowledge_base(json_data)
+                try:
+                    knowledge_base_uuid = await self.ap.knowledge_service.create_knowledge_base(json_data)
+                except ValueError as e:
+                    return self.http_status(400, -1, str(e))
                 return self.success(data={'uuid': knowledge_base_uuid})
 
             return self.http_status(405, -1, 'Method not allowed')
@@ -39,7 +42,7 @@ class KnowledgeBaseRouterGroup(group.RouterGroup):
             elif quart.request.method == 'PUT':
                 json_data = await quart.request.json
                 await self.ap.knowledge_service.update_knowledge_base(knowledge_base_uuid, json_data)
-                return self.success({})
+                return self.success(data={'uuid': knowledge_base_uuid})
 
             elif quart.request.method == 'DELETE':
                 await self.ap.knowledge_service.delete_knowledge_base(knowledge_base_uuid)
@@ -65,8 +68,12 @@ class KnowledgeBaseRouterGroup(group.RouterGroup):
                 if not file_id:
                     return self.http_status(400, -1, 'File ID is required')
 
+                parser_plugin_id = json_data.get('parser_plugin_id')
+
                 # 调用服务层方法将文件与知识库关联
-                task_id = await self.ap.knowledge_service.store_file(knowledge_base_uuid, file_id)
+                task_id = await self.ap.knowledge_service.store_file(
+                    knowledge_base_uuid, file_id, parser_plugin_id=parser_plugin_id
+                )
                 return self.success(
                     {
                         'task_id': task_id,
@@ -90,5 +97,13 @@ class KnowledgeBaseRouterGroup(group.RouterGroup):
         async def retrieve_knowledge_base(knowledge_base_uuid: str) -> str:
             json_data = await quart.request.json
             query = json_data.get('query')
-            results = await self.ap.knowledge_service.retrieve_knowledge_base(knowledge_base_uuid, query)
+
+            if not query or not query.strip():
+                return self.http_status(400, -1, 'Query is required and cannot be empty')
+
+            # Extract retrieval_settings to allow dynamic control over Knowledge Engine behavior (e.g. top_k, filters)
+            retrieval_settings = json_data.get('retrieval_settings', {})
+            results = await self.ap.knowledge_service.retrieve_knowledge_base(
+                knowledge_base_uuid, query, retrieval_settings
+            )
             return self.success(data={'results': results})

src/langbot/pkg/api/http/controller/groups/knowledge/engines.py

@@ -0,0 +1,45 @@
+import quart
+from urllib.parse import unquote
+from ... import group
+
+
+@group.group_class('knowledge_engines', '/api/v1/knowledge/engines')
+class KnowledgeEnginesRouterGroup(group.RouterGroup):
+    async def initialize(self) -> None:
+        @self.route('', methods=['GET'], auth_type=group.AuthType.USER_TOKEN_OR_API_KEY)
+        async def list_knowledge_engines() -> quart.Response:
+            """List all available Knowledge Engines from plugins.
+
+            Returns a list of Knowledge Engines with their capabilities and configuration schemas.
+            This is used by the frontend to render the knowledge base creation wizard.
+            """
+            engines = await self.ap.knowledge_service.list_knowledge_engines()
+            return self.success(data={'engines': engines})
+
+        @self.route(
+            '/<path:plugin_id>/creation-schema', methods=['GET'], auth_type=group.AuthType.USER_TOKEN_OR_API_KEY
+        )
+        async def get_engine_creation_schema(plugin_id: str) -> quart.Response:
+            """Get creation settings schema for a specific Knowledge Engine.
+
+            plugin_id is in 'author/name' format, captured via <path:> converter.
+            """
+            plugin_id = unquote(plugin_id)
+            if '/' not in plugin_id:
+                return self.http_status(400, -1, 'Invalid plugin_id format. Expected author/name.')
+            schema = await self.ap.knowledge_service.get_engine_creation_schema(plugin_id)
+            return self.success(data={'schema': schema})
+
+        @self.route(
+            '/<path:plugin_id>/retrieval-schema', methods=['GET'], auth_type=group.AuthType.USER_TOKEN_OR_API_KEY
+        )
+        async def get_engine_retrieval_schema(plugin_id: str) -> quart.Response:
+            """Get retrieval settings schema for a specific Knowledge Engine.
+
+            plugin_id is in 'author/name' format, captured via <path:> converter.
+            """
+            plugin_id = unquote(plugin_id)
+            if '/' not in plugin_id:
+                return self.http_status(400, -1, 'Invalid plugin_id format. Expected author/name.')
+            schema = await self.ap.knowledge_service.get_engine_retrieval_schema(plugin_id)
+            return self.success(data={'schema': schema})

src/langbot/pkg/api/http/controller/groups/knowledge/external.py

@@ -1,61 +0,0 @@
-import quart
-from ... import group
-
-
-@group.group_class('external_knowledge_base', '/api/v1/knowledge/external-bases')
-class ExternalKnowledgeBaseRouterGroup(group.RouterGroup):
-    async def initialize(self) -> None:
-        @self.route('/retrievers', methods=['GET'])
-        async def list_knowledge_retrievers() -> quart.Response:
-            """List all available knowledge retrievers from plugins."""
-            retrievers = await self.ap.plugin_connector.list_knowledge_retrievers()
-            return self.success(data={'retrievers': retrievers})
-
-        @self.route('', methods=['POST', 'GET'])
-        async def handle_external_knowledge_bases() -> quart.Response:
-            if quart.request.method == 'GET':
-                external_kbs = await self.ap.external_kb_service.get_external_knowledge_bases()
-                return self.success(data={'bases': external_kbs})
-
-            elif quart.request.method == 'POST':
-                json_data = await quart.request.json
-                kb_uuid = await self.ap.external_kb_service.create_external_knowledge_base(json_data)
-                return self.success(data={'uuid': kb_uuid})
-
-            return self.http_status(405, -1, 'Method not allowed')
-
-        @self.route(
-            '/<kb_uuid>',
-            methods=['GET', 'DELETE', 'PUT'],
-        )
-        async def handle_specific_external_knowledge_base(kb_uuid: str) -> quart.Response:
-            if quart.request.method == 'GET':
-                external_kb = await self.ap.external_kb_service.get_external_knowledge_base(kb_uuid)
-
-                if external_kb is None:
-                    return self.http_status(404, -1, 'external knowledge base not found')
-
-                return self.success(
-                    data={
-                        'base': external_kb,
-                    }
-                )
-
-            elif quart.request.method == 'PUT':
-                json_data = await quart.request.json
-                await self.ap.external_kb_service.update_external_knowledge_base(kb_uuid, json_data)
-                return self.success({})
-
-            elif quart.request.method == 'DELETE':
-                await self.ap.external_kb_service.delete_external_knowledge_base(kb_uuid)
-                return self.success({})
-
-        @self.route(
-            '/<kb_uuid>/retrieve',
-            methods=['POST'],
-        )
-        async def retrieve_external_knowledge_base(kb_uuid: str) -> str:
-            json_data = await quart.request.json
-            query = json_data.get('query')
-            results = await self.ap.external_kb_service.retrieve_external_knowledge_base(kb_uuid, query)
-            return self.success(data={'results': results})

src/langbot/pkg/api/http/controller/groups/knowledge/migration.py

@@ -0,0 +1,372 @@
+import asyncio
+import json
+
+import httpx
+import quart
+import sqlalchemy
+
+from ... import group
+from ......core import taskmgr
+from ......entity.persistence import metadata as persistence_metadata
+from langbot_plugin.runtime.plugin.mgr import PluginInstallSource
+
+LANGRAG_PLUGIN_AUTHOR = 'langbot-team'
+LANGRAG_PLUGIN_NAME = 'LangRAG'
+LANGRAG_PLUGIN_ID = f'{LANGRAG_PLUGIN_AUTHOR}/{LANGRAG_PLUGIN_NAME}'
+DEFAULT_SPACE_URL = 'https://space.langbot.app'
+
+# Old Retriever plugin_name -> New Connector plugin_name
+EXTERNAL_PLUGIN_NAME_MAPPING = {
+    'DifyDatasetsRetriever': 'DifyDatasetsConnector',
+    'RAGFlowRetriever': 'RAGFlowConnector',
+    'FastGPTRetriever': 'FastGPTConnector',
+}
+
+# Per-plugin: which old retriever_config fields belong to creation_settings.
+# Remaining fields go to retrieval_settings.
+# None means ALL fields go to creation_settings (no retrieval_schema).
+EXTERNAL_PLUGIN_CREATION_FIELDS: dict[str, set[str] | None] = {
+    'langbot-team/DifyDatasetsConnector': {'api_base_url', 'dify_apikey', 'dataset_id'},
+    'langbot-team/RAGFlowConnector': {'api_base_url', 'api_key', 'dataset_ids'},
+    'langbot-team/FastGPTConnector': None,  # all fields -> creation_settings
+}
+
+
+@group.group_class('knowledge/migration', '/api/v1/knowledge/migration')
+class KnowledgeMigrationRouterGroup(group.RouterGroup):
+    async def _get_migration_flag(self) -> bool:
+        """Check if rag_plugin_migration_needed flag is set."""
+        result = await self.ap.persistence_mgr.execute_async(
+            sqlalchemy.select(persistence_metadata.Metadata).where(
+                persistence_metadata.Metadata.key == 'rag_plugin_migration_needed'
+            )
+        )
+        row = result.first()
+        return row is not None and row.value == 'true'
+
+    async def _set_migration_flag(self, value: str):
+        """Set rag_plugin_migration_needed flag."""
+        await self.ap.persistence_mgr.execute_async(
+            sqlalchemy.update(persistence_metadata.Metadata)
+            .where(persistence_metadata.Metadata.key == 'rag_plugin_migration_needed')
+            .values(value=value)
+        )
+
+    async def _table_exists(self, table_name: str) -> bool:
+        """Check if a table exists."""
+        if self.ap.persistence_mgr.db.name == 'postgresql':
+            result = await self.ap.persistence_mgr.execute_async(
+                sqlalchemy.text(
+                    'SELECT EXISTS (SELECT FROM information_schema.tables WHERE table_name = :table_name);'
+                ).bindparams(table_name=table_name)
+            )
+            return result.scalar()
+        else:
+            result = await self.ap.persistence_mgr.execute_async(
+                sqlalchemy.text("SELECT name FROM sqlite_master WHERE type='table' AND name=:table_name;").bindparams(
+                    table_name=table_name
+                )
+            )
+            return result.first() is not None
+
+    async def _install_plugin_from_marketplace(
+        self, plugin_id: str, task_context: taskmgr.TaskContext, space_url: str
+    ) -> None:
+        """Install a single plugin from the marketplace."""
+        p_author, p_name = plugin_id.split('/', 1)
+        self.ap.logger.info(f'RAG migration: installing plugin {plugin_id} from marketplace...')
+        task_context.trace(f'Installing plugin {plugin_id} from marketplace...')
+
+        async with httpx.AsyncClient(trust_env=True, timeout=15) as client:
+            resp = await client.get(f'{space_url}/api/v1/marketplace/plugins/{p_author}/{p_name}')
+            resp.raise_for_status()
+            p_data = resp.json().get('data', {}).get('plugin', {})
+            p_version = p_data.get('latest_version')
+            if not p_version:
+                raise Exception(f'Could not determine latest version for {plugin_id}')
+
+        await self.ap.plugin_connector.install_plugin(
+            PluginInstallSource.MARKETPLACE,
+            {
+                'plugin_author': p_author,
+                'plugin_name': p_name,
+                'plugin_version': p_version,
+            },
+            task_context=task_context,
+        )
+        self.ap.logger.info(f'RAG migration: plugin {plugin_id} install request sent.')
+
+    async def _execute_rag_migration(self, task_context: taskmgr.TaskContext, install_plugin: bool = True):
+        """Execute RAG migration: install required plugins and restore backup data."""
+        warnings = []
+
+        # Collect all plugins we need: LangRAG (always) + connector plugins (from external KBs)
+        needed_plugins: dict[str, str] = {
+            LANGRAG_PLUGIN_ID: LANGRAG_PLUGIN_NAME,
+        }
+
+        has_external = await self._table_exists('external_knowledge_bases')
+        if has_external:
+            result = await self.ap.persistence_mgr.execute_async(
+                sqlalchemy.text('SELECT DISTINCT plugin_author, plugin_name FROM external_knowledge_bases;')
+            )
+            for row in result.fetchall():
+                plugin_author = row[0] or ''
+                plugin_name = row[1] or ''
+                mapped_name = EXTERNAL_PLUGIN_NAME_MAPPING.get(plugin_name, plugin_name)
+                plugin_id = f'{plugin_author}/{mapped_name}'
+                if plugin_id not in needed_plugins:
+                    needed_plugins[plugin_id] = mapped_name
+
+        self.ap.logger.info(f'RAG migration: plugins needed: {list(needed_plugins.keys())}')
+
+        if install_plugin:
+            # Step 1: Install all required plugins from marketplace
+            task_context.trace('Installing required plugins...', action='install-plugin')
+            space_url = self.ap.instance_config.data.get('space', {}).get('url', DEFAULT_SPACE_URL).rstrip('/')
+
+            for plugin_id in needed_plugins:
+                try:
+                    await self._install_plugin_from_marketplace(plugin_id, task_context, space_url)
+                except Exception as e:
+                    self.ap.logger.warning(f'RAG migration: plugin {plugin_id} install returned: {e}')
+                    task_context.trace(f'Plugin install note ({plugin_id}): {e}')
+
+            # Step 2: Wait for all plugins to become available as knowledge engines
+            task_context.trace(
+                f'Waiting for plugins to become available: {list(needed_plugins.keys())}...',
+                action='wait-plugin',
+            )
+            max_retries = 30
+            engine_id_set: set[str] = set()
+            for i in range(max_retries):
+                try:
+                    engines = await self.ap.plugin_connector.list_knowledge_engines()
+                    engine_id_set = {e.get('plugin_id') for e in engines}
+                except Exception:
+                    pass
+                if all(pid in engine_id_set for pid in needed_plugins):
+                    self.ap.logger.info(f'RAG migration: all plugins ready: {engine_id_set}')
+                    task_context.trace('All required plugins are ready.')
+                    break
+                if i == max_retries - 1:
+                    still_missing = [pid for pid in needed_plugins if pid not in engine_id_set]
+                    warning = f'Plugin(s) {still_missing} did not become available after {max_retries} retries'
+                    self.ap.logger.warning(f'RAG migration: {warning}')
+                    warnings.append(warning)
+                    task_context.trace(warning)
+                await asyncio.sleep(2)
+        else:
+            try:
+                engines = await self.ap.plugin_connector.list_knowledge_engines()
+                engine_id_set = {e.get('plugin_id') for e in engines}
+            except Exception:
+                engine_id_set = set()
+
+        # Step 3: Restore internal knowledge bases from backup
+        task_context.trace('Restoring internal knowledge bases...', action='restore-internal')
+        if await self._table_exists('knowledge_bases_backup'):
+            result = await self.ap.persistence_mgr.execute_async(
+                sqlalchemy.text('SELECT * FROM knowledge_bases_backup;')
+            )
+            rows = result.fetchall()
+            columns = result.keys()
+
+            for row in rows:
+                row_dict = dict(zip(columns, row))
+                kb_uuid = row_dict.get('uuid')
+                name = row_dict.get('name', 'Untitled')
+                description = row_dict.get('description', '')
+                emoji = row_dict.get('emoji', '\U0001f4da')
+                embedding_model_uuid = row_dict.get('embedding_model_uuid', '')
+                top_k = row_dict.get('top_k', 5)
+                created_at = row_dict.get('created_at')
+                updated_at = row_dict.get('updated_at')
+
+                creation_settings = json.dumps({'embedding_model_uuid': embedding_model_uuid})
+                retrieval_settings = json.dumps({'top_k': top_k})
+
+                await self.ap.persistence_mgr.execute_async(
+                    sqlalchemy.text(
+                        'INSERT INTO knowledge_bases '
+                        '(uuid, name, description, emoji, created_at, updated_at, '
+                        'knowledge_engine_plugin_id, collection_id, creation_settings, retrieval_settings) '
+                        'VALUES (:uuid, :name, :description, :emoji, :created_at, :updated_at, '
+                        ':plugin_id, :collection_id, :creation_settings, :retrieval_settings);'
+                    ).bindparams(
+                        uuid=kb_uuid,
+                        name=name,
+                        description=description,
+                        emoji=emoji,
+                        created_at=created_at,
+                        updated_at=updated_at,
+                        plugin_id=LANGRAG_PLUGIN_ID,
+                        collection_id=kb_uuid,
+                        creation_settings=creation_settings,
+                        retrieval_settings=retrieval_settings,
+                    )
+                )
+
+                try:
+                    config = {'embedding_model_uuid': embedding_model_uuid}
+                    await self.ap.plugin_connector.rag_on_kb_create(LANGRAG_PLUGIN_ID, kb_uuid, config)
+                    task_context.trace(f'Restored internal KB: {name} ({kb_uuid})')
+                except Exception as e:
+                    warning = f'Failed to notify plugin for KB {name} ({kb_uuid}): {e}'
+                    warnings.append(warning)
+                    task_context.trace(warning)
+
+            await self.ap.rag_mgr.load_knowledge_bases_from_db()
+
+        # Step 4: Restore external knowledge bases
+        task_context.trace('Restoring external knowledge bases...', action='restore-external')
+        if has_external:
+            result = await self.ap.persistence_mgr.execute_async(
+                sqlalchemy.text('SELECT * FROM external_knowledge_bases;')
+            )
+            rows = result.fetchall()
+            columns = result.keys()
+
+            self.ap.logger.info(
+                f'RAG migration: {len(rows)} external KB(s) to restore. Available engines: {engine_id_set}'
+            )
+            task_context.trace(f'Found {len(rows)} external KB(s). Available engines: {engine_id_set}')
+
+            for row in rows:
+                row_dict = dict(zip(columns, row))
+                kb_uuid = row_dict.get('uuid')
+                name = row_dict.get('name', 'Untitled')
+                description = row_dict.get('description', '')
+                emoji = row_dict.get('emoji', '\U0001f517')
+                plugin_author = row_dict.get('plugin_author', '')
+                plugin_name = row_dict.get('plugin_name', '')
+                retriever_config = row_dict.get('retriever_config', {})
+                created_at = row_dict.get('created_at')
+
+                mapped_plugin_name = EXTERNAL_PLUGIN_NAME_MAPPING.get(plugin_name, plugin_name)
+                external_plugin_id = f'{plugin_author}/{mapped_plugin_name}'
+
+                self.ap.logger.info(
+                    f'RAG migration: processing external KB "{name}" ({kb_uuid}), '
+                    f'plugin: {plugin_author}/{plugin_name} -> {external_plugin_id}'
+                )
+
+                if isinstance(retriever_config, str):
+                    try:
+                        retriever_config = json.loads(retriever_config)
+                    except (json.JSONDecodeError, TypeError):
+                        retriever_config = {}
+
+                creation_fields = EXTERNAL_PLUGIN_CREATION_FIELDS.get(external_plugin_id)
+                if creation_fields is None:
+                    creation_settings_dict = retriever_config
+                    retrieval_settings_dict = {}
+                else:
+                    creation_settings_dict = {k: v for k, v in retriever_config.items() if k in creation_fields}
+                    retrieval_settings_dict = {k: v for k, v in retriever_config.items() if k not in creation_fields}
+
+                await self.ap.persistence_mgr.execute_async(
+                    sqlalchemy.text(
+                        'INSERT INTO knowledge_bases '
+                        '(uuid, name, description, emoji, created_at, updated_at, '
+                        'knowledge_engine_plugin_id, collection_id, creation_settings, retrieval_settings) '
+                        'VALUES (:uuid, :name, :description, :emoji, :created_at, :updated_at, '
+                        ':plugin_id, :collection_id, :creation_settings, :retrieval_settings);'
+                    ).bindparams(
+                        uuid=kb_uuid,
+                        name=name,
+                        description=description,
+                        emoji=emoji,
+                        created_at=created_at,
+                        updated_at=created_at,
+                        plugin_id=external_plugin_id,
+                        collection_id=kb_uuid,
+                        creation_settings=json.dumps(creation_settings_dict),
+                        retrieval_settings=json.dumps(retrieval_settings_dict),
+                    )
+                )
+
+                if external_plugin_id not in engine_id_set:
+                    warning = (
+                        f'External KB "{name}" ({kb_uuid}) record saved, but plugin {external_plugin_id} '
+                        f'is not installed yet. Install the connector plugin to use it.'
+                    )
+                    warnings.append(warning)
+                    task_context.trace(warning)
+                else:
+                    try:
+                        await self.ap.plugin_connector.rag_on_kb_create(
+                            external_plugin_id, kb_uuid, creation_settings_dict
+                        )
+                        task_context.trace(f'Restored external KB: {name} ({kb_uuid})')
+                    except Exception as e:
+                        warning = f'Failed to notify plugin for external KB {name} ({kb_uuid}): {e}'
+                        warnings.append(warning)
+                        task_context.trace(warning)
+
+            await self.ap.rag_mgr.load_knowledge_bases_from_db()
+
+        # Step 5: Clear migration flag
+        await self._set_migration_flag('false')
+        task_context.trace('RAG migration completed.', action='done')
+
+        if warnings:
+            task_context.trace(f'Completed with {len(warnings)} warning(s).')
+
+    async def initialize(self) -> None:
+        @self.route('/status', methods=['GET'], auth_type=group.AuthType.USER_TOKEN)
+        async def _() -> str:
+            needed = await self._get_migration_flag()
+
+            internal_kb_count = 0
+            external_kb_count = 0
+
+            if needed:
+                if await self._table_exists('knowledge_bases_backup'):
+                    result = await self.ap.persistence_mgr.execute_async(
+                        sqlalchemy.text('SELECT COUNT(*) FROM knowledge_bases_backup;')
+                    )
+                    internal_kb_count = result.scalar() or 0
+
+                if await self._table_exists('external_knowledge_bases'):
+                    result = await self.ap.persistence_mgr.execute_async(
+                        sqlalchemy.text('SELECT COUNT(*) FROM external_knowledge_bases;')
+                    )
+                    external_kb_count = result.scalar() or 0
+
+            return self.success(
+                data={
+                    'needed': needed,
+                    'internal_kb_count': internal_kb_count,
+                    'external_kb_count': external_kb_count,
+                }
+            )
+
+        @self.route('/execute', methods=['POST'], auth_type=group.AuthType.USER_TOKEN)
+        async def _() -> str:
+            needed = await self._get_migration_flag()
+            if not needed:
+                return self.http_status(400, -1, 'RAG migration is not needed')
+
+            data = await quart.request.get_json(silent=True) or {}
+            install_plugin = data.get('install_plugin', True)
+
+            ctx = taskmgr.TaskContext.new()
+            wrapper = self.ap.task_mgr.create_user_task(
+                self._execute_rag_migration(task_context=ctx, install_plugin=install_plugin),
+                kind='rag-migration',
+                name='rag-migration-execute',
+                label='Migrating knowledge bases to plugin architecture',
+                context=ctx,
+            )
+
+            return self.success(data={'task_id': wrapper.id})
+
+        @self.route('/dismiss', methods=['POST'], auth_type=group.AuthType.USER_TOKEN)
+        async def _() -> str:
+            needed = await self._get_migration_flag()
+            if not needed:
+                return self.http_status(400, -1, 'RAG migration is not needed')
+
+            await self._set_migration_flag('false')
+            return self.success()

src/langbot/pkg/api/http/controller/groups/knowledge/parsers.py

@@ -0,0 +1,16 @@
+import quart
+from ... import group
+
+
+@group.group_class('parsers', '/api/v1/knowledge/parsers')
+class ParsersRouterGroup(group.RouterGroup):
+    async def initialize(self) -> None:
+        @self.route('', methods=['GET'], auth_type=group.AuthType.USER_TOKEN_OR_API_KEY)
+        async def list_parsers() -> quart.Response:
+            """List all available parsers from plugins.
+
+            Optional query parameter `mime_type` to filter parsers by supported MIME type.
+            """
+            mime_type = quart.request.args.get('mime_type')
+            parsers = await self.ap.knowledge_service.list_parsers(mime_type)
+            return self.success(data={'parsers': parsers})

src/langbot/pkg/api/http/controller/groups/pipelines/pipelines.py

@@ -68,7 +68,7 @@ class PipelinesRouterGroup(group.RouterGroup):
                     return self.http_status(404, -1, 'pipeline not found')
 
                 # Only include plugins with pipeline-related components (Command, EventListener, Tool)
-                # Plugins that only have KnowledgeRetriever components are not suitable for pipeline extensions
+                # Plugins that only have KnowledgeEngine components are not suitable for pipeline extensions
                 pipeline_component_kinds = ['Command', 'EventListener', 'Tool']
                 plugins = await self.ap.plugin_connector.list_plugins(component_kinds=pipeline_component_kinds)
                 mcp_servers = await self.ap.mcp_service.get_mcp_servers(contain_runtime_info=True)

src/langbot/pkg/api/http/service/bot.py

@@ -70,12 +70,17 @@ class BotService:
             'lark',
         ]:
             webhook_prefix = self.ap.instance_config.data['api'].get('webhook_prefix', 'http://127.0.0.1:5300')
+            extra_webhook_prefix = self.ap.instance_config.data['api'].get('extra_webhook_prefix', '')
             webhook_url = f'/bots/{bot_uuid}'
             adapter_runtime_values['webhook_url'] = webhook_url
             adapter_runtime_values['webhook_full_url'] = f'{webhook_prefix}{webhook_url}'
+            adapter_runtime_values['extra_webhook_full_url'] = (
+                f'{extra_webhook_prefix}{webhook_url}' if extra_webhook_prefix else ''
+            )
         else:
             adapter_runtime_values['webhook_url'] = None
             adapter_runtime_values['webhook_full_url'] = None
+            adapter_runtime_values['extra_webhook_full_url'] = None
 
         persistence_bot['adapter_runtime_values'] = adapter_runtime_values
 

src/langbot/pkg/api/http/service/external_kb.py

@@ -1,80 +0,0 @@
-from __future__ import annotations
-
-from ....core import app
-import sqlalchemy
-from langbot.pkg.entity.persistence import rag as persistence_rag
-import uuid
-
-
-class ExternalKBService:
-    """External KB service"""
-
-    ap: app.Application
-
-    def __init__(self, ap: app.Application) -> None:
-        self.ap = ap
-
-    # External Knowledge Base methods
-    async def get_external_knowledge_bases(self) -> list[dict]:
-        result = await self.ap.persistence_mgr.execute_async(sqlalchemy.select(persistence_rag.ExternalKnowledgeBase))
-        external_kbs = result.all()
-        return [
-            self.ap.persistence_mgr.serialize_model(persistence_rag.ExternalKnowledgeBase, external_kb)
-            for external_kb in external_kbs
-        ]
-
-    async def get_external_knowledge_base(self, kb_uuid: str) -> dict | None:
-        result = await self.ap.persistence_mgr.execute_async(
-            sqlalchemy.select(persistence_rag.ExternalKnowledgeBase).where(
-                persistence_rag.ExternalKnowledgeBase.uuid == kb_uuid
-            )
-        )
-        external_kb = result.first()
-        if external_kb is None:
-            return None
-        return self.ap.persistence_mgr.serialize_model(persistence_rag.ExternalKnowledgeBase, external_kb)
-
-    async def create_external_knowledge_base(self, kb_data: dict) -> str:
-        kb_data['uuid'] = str(uuid.uuid4())
-        await self.ap.persistence_mgr.execute_async(
-            sqlalchemy.insert(persistence_rag.ExternalKnowledgeBase).values(kb_data)
-        )
-
-        kb = await self.get_external_knowledge_base(kb_data['uuid'])
-
-        await self.ap.rag_mgr.load_external_knowledge_base(kb)
-
-        return kb_data['uuid']
-
-    async def retrieve_external_knowledge_base(self, kb_uuid: str, query: str) -> list[dict]:
-        """Retrieve external knowledge base"""
-        runtime_kb = await self.ap.rag_mgr.get_knowledge_base_by_uuid(kb_uuid)
-        if runtime_kb is None:
-            raise Exception('Knowledge base not found')
-        return [
-            result.model_dump() for result in await runtime_kb.retrieve(query, 5)
-        ]  # top_k is just a placeholder for external knowledge base
-
-    async def update_external_knowledge_base(self, kb_uuid: str, kb_data: dict) -> None:
-        if 'uuid' in kb_data:
-            del kb_data['uuid']
-
-        await self.ap.persistence_mgr.execute_async(
-            sqlalchemy.update(persistence_rag.ExternalKnowledgeBase)
-            .values(kb_data)
-            .where(persistence_rag.ExternalKnowledgeBase.uuid == kb_uuid)
-        )
-        await self.ap.rag_mgr.remove_knowledge_base_from_runtime(kb_uuid)
-
-        kb = await self.get_external_knowledge_base(kb_uuid)
-
-        await self.ap.rag_mgr.load_external_knowledge_base(kb)
-
-    async def delete_external_knowledge_base(self, kb_uuid: str) -> None:
-        await self.ap.persistence_mgr.execute_async(
-            sqlalchemy.delete(persistence_rag.ExternalKnowledgeBase).where(
-                persistence_rag.ExternalKnowledgeBase.uuid == kb_uuid
-            )
-        )
-
-        await self.ap.rag_mgr.delete_knowledge_base(kb_uuid)

src/langbot/pkg/api/http/service/knowledge.py

@@ -1,6 +1,5 @@
 from __future__ import annotations
 
-import uuid
 import sqlalchemy
 
 from ....core import app
@@ -17,64 +16,77 @@ class KnowledgeService:
 
     async def get_knowledge_bases(self) -> list[dict]:
         """获取所有知识库"""
-        result = await self.ap.persistence_mgr.execute_async(sqlalchemy.select(persistence_rag.KnowledgeBase))
-        knowledge_bases = result.all()
-        return [
-            self.ap.persistence_mgr.serialize_model(persistence_rag.KnowledgeBase, knowledge_base)
-            for knowledge_base in knowledge_bases
-        ]
+        return await self.ap.rag_mgr.get_all_knowledge_base_details()
 
     async def get_knowledge_base(self, kb_uuid: str) -> dict | None:
         """获取知识库"""
-        result = await self.ap.persistence_mgr.execute_async(
-            sqlalchemy.select(persistence_rag.KnowledgeBase).where(persistence_rag.KnowledgeBase.uuid == kb_uuid)
-        )
-        knowledge_base = result.first()
-        if knowledge_base is None:
-            return None
-        return self.ap.persistence_mgr.serialize_model(persistence_rag.KnowledgeBase, knowledge_base)
+        return await self.ap.rag_mgr.get_knowledge_base_details(kb_uuid)
 
     async def create_knowledge_base(self, kb_data: dict) -> str:
         """创建知识库"""
-        kb_data['uuid'] = str(uuid.uuid4())
-        await self.ap.persistence_mgr.execute_async(sqlalchemy.insert(persistence_rag.KnowledgeBase).values(kb_data))
+        # In new architecture, we delegate entirely to RAGManager which uses plugins.
+        # Legacy internal KB creation is removed.
 
-        kb = await self.get_knowledge_base(kb_data['uuid'])
+        knowledge_engine_plugin_id = kb_data.get('knowledge_engine_plugin_id')
+        if not knowledge_engine_plugin_id:
+            raise ValueError('knowledge_engine_plugin_id is required')
 
-        await self.ap.rag_mgr.load_knowledge_base(kb)
-
-        return kb_data['uuid']
+        kb = await self.ap.rag_mgr.create_knowledge_base(
+            name=kb_data.get('name', 'Untitled'),
+            knowledge_engine_plugin_id=knowledge_engine_plugin_id,
+            creation_settings=kb_data.get('creation_settings', {}),
+            retrieval_settings=kb_data.get('retrieval_settings', {}),
+            description=kb_data.get('description', ''),
+        )
+        return kb.uuid
 
     async def update_knowledge_base(self, kb_uuid: str, kb_data: dict) -> None:
         """更新知识库"""
-        if 'uuid' in kb_data:
-            del kb_data['uuid']
+        # Filter to only mutable fields
+        filtered_data = {k: v for k, v in kb_data.items() if k in persistence_rag.KnowledgeBase.MUTABLE_FIELDS}
 
-        if 'embedding_model_uuid' in kb_data:
-            del kb_data['embedding_model_uuid']
+        if not filtered_data:
+            return
 
         await self.ap.persistence_mgr.execute_async(
             sqlalchemy.update(persistence_rag.KnowledgeBase)
-            .values(kb_data)
+            .values(filtered_data)
             .where(persistence_rag.KnowledgeBase.uuid == kb_uuid)
         )
         await self.ap.rag_mgr.remove_knowledge_base_from_runtime(kb_uuid)
 
         kb = await self.get_knowledge_base(kb_uuid)
+        if kb is None:
+            raise Exception('Knowledge base not found after update')
 
         await self.ap.rag_mgr.load_knowledge_base(kb)
 
-    async def store_file(self, kb_uuid: str, file_id: str) -> int:
+    async def _check_doc_capability(self, kb_uuid: str, operation: str) -> None:
+        """Check if the KB's Knowledge Engine supports document operations.
+
+        Args:
+            kb_uuid: Knowledge base UUID.
+            operation: Human-readable operation name for error messages.
+
+        Raises:
+            Exception: If the KB does not support doc_ingestion.
+        """
+        kb_info = await self.ap.rag_mgr.get_knowledge_base_details(kb_uuid)
+        if not kb_info:
+            raise Exception('Knowledge base not found')
+        capabilities = kb_info.get('knowledge_engine', {}).get('capabilities', [])
+        if 'doc_ingestion' not in capabilities:
+            raise Exception(f'This knowledge base does not support {operation}')
+
+    async def store_file(self, kb_uuid: str, file_id: str, parser_plugin_id: str | None = None) -> str:
         """存储文件"""
-        # await self.ap.persistence_mgr.execute_async(sqlalchemy.insert(persistence_rag.File).values(kb_id=kb_uuid, file_id=file_id))
-        # await self.ap.rag_mgr.store_file(file_id)
         runtime_kb = await self.ap.rag_mgr.get_knowledge_base_by_uuid(kb_uuid)
         if runtime_kb is None:
             raise Exception('Knowledge base not found')
-        # Only internal KBs support file storage
-        if runtime_kb.get_type() != 'internal':
-            raise Exception('Only internal knowledge bases support file storage')
-        result = await runtime_kb.store_file(file_id)
+
+        await self._check_doc_capability(kb_uuid, 'document upload')
+
+        result = await runtime_kb.store_file(file_id, parser_plugin_id=parser_plugin_id)
 
         # Update the KB's updated_at timestamp
         await self.ap.persistence_mgr.execute_async(
@@ -85,14 +97,18 @@ class KnowledgeService:
 
         return result
 
-    async def retrieve_knowledge_base(self, kb_uuid: str, query: str) -> list[dict]:
+    async def retrieve_knowledge_base(
+        self, kb_uuid: str, query: str, retrieval_settings: dict | None = None
+    ) -> list[dict]:
         """检索知识库"""
         runtime_kb = await self.ap.rag_mgr.get_knowledge_base_by_uuid(kb_uuid)
         if runtime_kb is None:
             raise Exception('Knowledge base not found')
-        return [
-            result.model_dump() for result in await runtime_kb.retrieve(query, runtime_kb.knowledge_base_entity.top_k)
-        ]
+
+        # Pass retrieval_settings
+        results = await runtime_kb.retrieve(query, settings=retrieval_settings)
+
+        return [result.model_dump() for result in results]
 
     async def get_files_by_knowledge_base(self, kb_uuid: str) -> list[dict]:
         """获取知识库文件"""
@@ -107,9 +123,9 @@ class KnowledgeService:
         runtime_kb = await self.ap.rag_mgr.get_knowledge_base_by_uuid(kb_uuid)
         if runtime_kb is None:
             raise Exception('Knowledge base not found')
-        # Only internal KBs support file deletion
-        if runtime_kb.get_type() != 'internal':
-            raise Exception('Only internal knowledge bases support file deletion')
+
+        await self._check_doc_capability(kb_uuid, 'document deletion')
+
         await runtime_kb.delete_file(file_id)
 
         # Update the KB's updated_at timestamp
@@ -121,13 +137,14 @@ class KnowledgeService:
 
     async def delete_knowledge_base(self, kb_uuid: str) -> None:
         """删除知识库"""
-        await self.ap.rag_mgr.delete_knowledge_base(kb_uuid)
-
+        # Delete from DB first to commit the deletion, then clean up runtime/plugin (best-effort)
         await self.ap.persistence_mgr.execute_async(
             sqlalchemy.delete(persistence_rag.KnowledgeBase).where(persistence_rag.KnowledgeBase.uuid == kb_uuid)
         )
 
         # delete files
+        # NOTE: Chunk cleanup is for legacy (pre-plugin) KBs that stored chunks locally.
+        # For plugin-based Knowledge Engines, the Chunk table is not populated, so this is a no-op.
         files = await self.ap.persistence_mgr.execute_async(
             sqlalchemy.select(persistence_rag.File).where(persistence_rag.File.kb_id == kb_uuid)
         )
@@ -140,3 +157,53 @@ class KnowledgeService:
             await self.ap.persistence_mgr.execute_async(
                 sqlalchemy.delete(persistence_rag.File).where(persistence_rag.File.uuid == file.uuid)
             )
+
+        # Remove from runtime and notify plugin (best-effort, DB is already cleaned up)
+        await self.ap.rag_mgr.delete_knowledge_base(kb_uuid)
+
+    # ================= Knowledge Engine Discovery =================
+
+    async def list_knowledge_engines(self) -> list[dict]:
+        """List all available Knowledge Engines from plugins."""
+        engines = []
+
+        if not self.ap.plugin_connector.is_enable_plugin:
+            return engines
+
+        # Get KnowledgeEngine plugins
+        try:
+            knowledge_engines = await self.ap.plugin_connector.list_knowledge_engines()
+            engines.extend(knowledge_engines)
+        except Exception as e:
+            self.ap.logger.warning(f'Failed to list Knowledge Engines from plugins: {e}')
+
+        return engines
+
+    async def list_parsers(self, mime_type: str | None = None) -> list[dict]:
+        """List available parsers, optionally filtered by MIME type."""
+        if not self.ap.plugin_connector.is_enable_plugin:
+            return []
+        try:
+            parsers = await self.ap.plugin_connector.list_parsers()
+            if mime_type:
+                parsers = [p for p in parsers if mime_type in p.get('supported_mime_types', [])]
+            return parsers
+        except Exception as e:
+            self.ap.logger.warning(f'Failed to list parsers: {e}')
+            return []
+
+    async def get_engine_creation_schema(self, plugin_id: str) -> dict:
+        """Get creation settings schema for a specific Knowledge Engine."""
+        try:
+            return await self.ap.plugin_connector.get_rag_creation_schema(plugin_id)
+        except Exception as e:
+            self.ap.logger.warning(f'Failed to get creation schema for {plugin_id}: {e}')
+            return {}
+
+    async def get_engine_retrieval_schema(self, plugin_id: str) -> dict:
+        """Get retrieval settings schema for a specific Knowledge Engine."""
+        try:
+            return await self.ap.plugin_connector.get_rag_retrieval_schema(plugin_id)
+        except Exception as e:
+            self.ap.logger.warning(f'Failed to get retrieval schema for {plugin_id}: {e}')
+            return {}

src/langbot/pkg/api/http/service/model.py

@@ -105,11 +105,16 @@ class LLMModelsService:
                 )
             )
             pipeline = result.first()
-            if pipeline is not None and pipeline.config['ai']['local-agent']['model'] == '':
-                pipeline_config = pipeline.config
-                pipeline_config['ai']['local-agent']['model'] = model_data['uuid']
-                pipeline_data = {'config': pipeline_config}
-                await self.ap.pipeline_service.update_pipeline(pipeline.uuid, pipeline_data)
+            if pipeline is not None:
+                model_config = pipeline.config.get('ai', {}).get('local-agent', {}).get('model', {})
+                if not model_config.get('primary', ''):
+                    pipeline_config = pipeline.config
+                    pipeline_config['ai']['local-agent']['model'] = {
+                        'primary': model_data['uuid'],
+                        'fallbacks': [],
+                    }
+                    pipeline_data = {'config': pipeline_config}
+                    await self.ap.pipeline_service.update_pipeline(pipeline.uuid, pipeline_data)
 
         return model_data['uuid']
 

src/langbot/pkg/api/http/service/monitoring.py

@@ -30,6 +30,7 @@ class MonitoringService:
         level: str = 'info',
         platform: str | None = None,
         user_id: str | None = None,
+        user_name: str | None = None,
         runner_name: str | None = None,
         variables: str | None = None,
         role: str = 'user',
@@ -49,6 +50,7 @@ class MonitoringService:
             'level': level,
             'platform': platform,
             'user_id': user_id,
+            'user_name': user_name,
             'runner_name': runner_name,
             'variables': variables,
             'role': role,
@@ -152,6 +154,7 @@ class MonitoringService:
         pipeline_name: str,
         platform: str | None = None,
         user_id: str | None = None,
+        user_name: str | None = None,
     ) -> None:
         """Record a new session"""
         session_data = {
@@ -166,6 +169,7 @@ class MonitoringService:
             'is_active': True,
             'platform': platform,
             'user_id': user_id,
+            'user_name': user_name,
         }
 
         await self.ap.persistence_mgr.execute_async(

src/langbot/pkg/api/http/service/space.py

@@ -1,6 +1,6 @@
 from __future__ import annotations
 
-import aiohttp
+from langbot.pkg.utils import httpclient
 import typing
 import datetime
 import time
@@ -99,49 +99,49 @@ class SpaceService:
         space_config = self._get_space_config()
         space_url = space_config['url']
 
-        async with aiohttp.ClientSession() as session:
-            async with session.post(
-                f'{space_url}/api/v1/accounts/oauth/token',
-                json={'code': code, 'instance_id': constants.instance_id},
-            ) as response:
-                if response.status != 200:
-                    raise ValueError(f'Failed to exchange OAuth code: {await response.text()}')
-                data = await response.json()
-                if data.get('code') != 0:
-                    raise ValueError(f'Failed to exchange OAuth code: {data.get("msg")}')
-                return data.get('data', {})
+        session = httpclient.get_session()
+        async with session.post(
+            f'{space_url}/api/v1/accounts/oauth/token',
+            json={'code': code, 'instance_id': constants.instance_id},
+        ) as response:
+            if response.status != 200:
+                raise ValueError(f'Failed to exchange OAuth code: {await response.text()}')
+            data = await response.json()
+            if data.get('code') != 0:
+                raise ValueError(f'Failed to exchange OAuth code: {data.get("msg")}')
+            return data.get('data', {})
 
     async def refresh_token(self, refresh_token: str) -> typing.Dict:
         """Refresh Space access token"""
         space_config = self._get_space_config()
         space_url = space_config['url']
 
-        async with aiohttp.ClientSession() as session:
-            async with session.post(
-                f'{space_url}/api/v1/accounts/token/refresh', json={'refresh_token': refresh_token}
-            ) as response:
-                if response.status != 200:
-                    raise ValueError(f'Failed to refresh token: {await response.text()}')
-                data = await response.json()
-                if data.get('code') != 0:
-                    raise ValueError(f'Failed to refresh token: {data.get("msg")}')
-                return data.get('data', {})
+        session = httpclient.get_session()
+        async with session.post(
+            f'{space_url}/api/v1/accounts/token/refresh', json={'refresh_token': refresh_token}
+        ) as response:
+            if response.status != 200:
+                raise ValueError(f'Failed to refresh token: {await response.text()}')
+            data = await response.json()
+            if data.get('code') != 0:
+                raise ValueError(f'Failed to refresh token: {data.get("msg")}')
+            return data.get('data', {})
 
     async def get_user_info_raw(self, access_token: str) -> typing.Dict:
         """Get user info from Space using access token (no validation)"""
         space_config = self._get_space_config()
         space_url = space_config['url']
 
-        async with aiohttp.ClientSession() as session:
-            async with session.get(
-                f'{space_url}/api/v1/accounts/me', headers={'Authorization': f'Bearer {access_token}'}
-            ) as response:
-                if response.status != 200:
-                    raise ValueError(f'Failed to get user info: {await response.text()}')
-                data = await response.json()
-                if data.get('code') != 0:
-                    raise ValueError(f'Failed to get user info: {data.get("msg")}')
-                return data.get('data', {})
+        session = httpclient.get_session()
+        async with session.get(
+            f'{space_url}/api/v1/accounts/me', headers={'Authorization': f'Bearer {access_token}'}
+        ) as response:
+            if response.status != 200:
+                raise ValueError(f'Failed to get user info: {await response.text()}')
+            data = await response.json()
+            if data.get('code') != 0:
+                raise ValueError(f'Failed to get user info: {data.get("msg")}')
+            return data.get('data', {})
 
     # === API calls with token validation ===
 
@@ -178,12 +178,12 @@ class SpaceService:
         space_config = self._get_space_config()
         space_url = space_config['url']
 
-        async with aiohttp.ClientSession() as session:
-            async with session.get(f'{space_url}/api/v1/models') as response:
-                if response.status != 200:
-                    raise ValueError(f'Failed to get models: {await response.text()}')
-                data = await response.json()
-                if data.get('code') != 0:
-                    raise ValueError(f'Failed to get models: {data.get("msg")}')
-                models_data = data.get('data', {}).get('models', [])
-                return [SpaceModel.model_validate(model_dict) for model_dict in models_data]
+        session = httpclient.get_session()
+        async with session.get(f'{space_url}/api/v1/models') as response:
+            if response.status != 200:
+                raise ValueError(f'Failed to get models: {await response.text()}')
+            data = await response.json()
+            if data.get('code') != 0:
+                raise ValueError(f'Failed to get models: {data.get("msg")}')
+            models_data = data.get('data', {}).get('models', [])
+            return [SpaceModel.model_validate(model_dict) for model_dict in models_data]

src/langbot/pkg/core/app.py

@@ -9,6 +9,7 @@ from ..platform import botmgr as im_mgr
 from ..platform.webhook_pusher import WebhookPusher
 from ..provider.session import sessionmgr as llm_session_mgr
 from ..provider.modelmgr import modelmgr as llm_model_mgr
+
 from langbot.pkg.provider.tools import toolmgr as llm_tool_mgr
 from ..config import manager as config_mgr
 from ..command import cmdmgr
@@ -29,14 +30,15 @@ from ..api.http.service import knowledge as knowledge_service
 from ..api.http.service import mcp as mcp_service
 from ..api.http.service import apikey as apikey_service
 from ..api.http.service import webhook as webhook_service
-from ..api.http.service import external_kb as external_kb_service
 from ..api.http.service import monitoring as monitoring_service
+
 from ..discover import engine as discover_engine
 from ..storage import mgr as storagemgr
 from ..utils import logcache
 from . import taskmgr
 from . import entities as core_entities
 from ..rag.knowledge import kbmgr as rag_mgr
+from ..rag.service import RAGRuntimeService
 from ..vector import mgr as vectordb_mgr
 from ..telemetry import telemetry as telemetry_module
 from ..survey import manager as survey_module
@@ -63,6 +65,7 @@ class Application:
     model_mgr: llm_model_mgr.ModelManager = None
 
     rag_mgr: rag_mgr.RAGManager = None
+    rag_runtime_service: RAGRuntimeService = None
 
     # TODO move to pipeline
     tool_mgr: llm_tool_mgr.ToolManager = None
@@ -138,8 +141,6 @@ class Application:
 
     knowledge_service: knowledge_service.KnowledgeService = None
 
-    external_kb_service: external_kb_service.ExternalKBService = None
-
     mcp_service: mcp_service.MCPService = None
 
     apikey_service: apikey_service.ApiKeyService = None

src/langbot/pkg/core/bootutils/deps.py

@@ -1,3 +1,4 @@
+import importlib.util
 import pip
 import os
 from ...utils import pkgmgr
@@ -49,9 +50,10 @@ async def check_deps() -> list[str]:
 
     missing_deps = []
     for dep in required_deps:
-        try:
-            __import__(dep)
-        except ImportError:
+        # Use find_spec instead of __import__ to avoid actually loading
+        # all modules into memory. find_spec only checks if the module
+        # can be found, without executing module-level code.
+        if importlib.util.find_spec(dep) is None:
             missing_deps.append(dep)
     return missing_deps
 

src/langbot/pkg/core/stages/build_app.py

@@ -12,6 +12,7 @@ from ...provider.session import sessionmgr as llm_session_mgr
 from ...provider.modelmgr import modelmgr as llm_model_mgr
 from ...provider.tools import toolmgr as llm_tool_mgr
 from ...rag.knowledge import kbmgr as rag_mgr
+from ...rag.service import RAGRuntimeService
 from ...platform import botmgr as im_mgr
 from ...platform.webhook_pusher import WebhookPusher
 from ...persistence import mgr as persistencemgr
@@ -26,7 +27,6 @@ from ...api.http.service import knowledge as knowledge_service
 from ...api.http.service import mcp as mcp_service
 from ...api.http.service import apikey as apikey_service
 from ...api.http.service import webhook as webhook_service
-from ...api.http.service import external_kb as external_kb_service
 from ...api.http.service import monitoring as monitoring_service
 from ...discover import engine as discover_engine
 from ...storage import mgr as storagemgr
@@ -73,9 +73,6 @@ class BuildAppStage(stage.BootingStage):
         knowledge_service_inst = knowledge_service.KnowledgeService(ap)
         ap.knowledge_service = knowledge_service_inst
 
-        external_kb_service_inst = external_kb_service.ExternalKBService(ap)
-        ap.external_kb_service = external_kb_service_inst
-
         mcp_service_inst = mcp_service.MCPService(ap)
         ap.mcp_service = mcp_service_inst
 
@@ -152,6 +149,9 @@ class BuildAppStage(stage.BootingStage):
         await rag_mgr_inst.initialize()
         ap.rag_mgr = rag_mgr_inst
 
+        # Initialize RAG Runtime Service for plugins
+        ap.rag_runtime_service = RAGRuntimeService(ap)
+
         # 初始化向量数据库管理器
         vectordb_mgr_inst = vectordb_mgr.VectorDBManager(ap)
         await vectordb_mgr_inst.initialize()

src/langbot/pkg/core/stages/load_config.py

@@ -74,20 +74,26 @@ def _apply_env_overrides_to_config(cfg: dict) -> dict:
         current = cfg
 
         for i, key in enumerate(keys):
-            if not isinstance(current, dict) or key not in current:
+            if not isinstance(current, dict):
                 break
 
             if i == len(keys) - 1:
-                # At the final key - check if it's a scalar value
-                if isinstance(current[key], (dict, list)):
-                    # Skip dict and list types
-                    pass
+                # At the final key
+                if key in current:
+                    if isinstance(current[key], (dict, list)):
+                        # Skip dict and list types
+                        pass
+                    else:
+                        # Valid scalar value - convert and set it
+                        converted_value = convert_value(env_value, current[key])
+                        current[key] = converted_value
                 else:
-                    # Valid scalar value - convert and set it
-                    converted_value = convert_value(env_value, current[key])
-                    current[key] = converted_value
+                    # Key doesn't exist yet - create it as string
+                    current[key] = env_value
             else:
-                # Navigate deeper
+                # Navigate deeper - create intermediate dict if needed
+                if key not in current:
+                    current[key] = {}
                 current = current[key]
 
     return cfg
@@ -146,16 +152,50 @@ class LoadConfigStage(stage.BootingStage):
         await ap.instance_config.dump_config()
 
         # load or generate instance id
-        ap.instance_id = await config.load_json_config(
-            'data/labels/instance_id.json',
-            template_data={
-                'instance_id': f'instance_{str(uuid.uuid4())}',
-                'instance_create_ts': int(time.time()),
-            },
-            completion=False,
-        )
+        # Priority:
+        # 1. system.instance_id from config.yaml (can be set via SYSTEM__INSTANCE_ID env var)
+        # 2. data/labels/instance_id.json (if file exists)
+        # 3. Generate new and save to file
+        config_instance_id = ap.instance_config.data.get('system', {}).get('instance_id', '')
 
-        constants.instance_id = ap.instance_id.data['instance_id']
+        if config_instance_id:
+            # Use the instance_id from config.yaml
+            constants.instance_id = config_instance_id
+            # Still load/create the file for backward compat, but don't use its value
+            ap.instance_id = await config.load_json_config(
+                'data/labels/instance_id.json',
+                template_data={
+                    'instance_id': f'instance_{str(uuid.uuid4())}',
+                    'instance_create_ts': int(time.time()),
+                },
+                completion=False,
+            )
+        else:
+            # Try loading file-based instance id
+            instance_id_path = os.path.join('data', 'labels', 'instance_id.json')
+            if os.path.exists(instance_id_path):
+                # File exists, read it
+                ap.instance_id = await config.load_json_config(
+                    'data/labels/instance_id.json',
+                    template_data={
+                        'instance_id': '',
+                        'instance_create_ts': 0,
+                    },
+                    completion=False,
+                )
+                constants.instance_id = ap.instance_id.data['instance_id']
+            else:
+                # Neither config nor file, generate new and save to file
+                new_id = f'instance_{str(uuid.uuid4())}'
+                ap.instance_id = await config.load_json_config(
+                    'data/labels/instance_id.json',
+                    template_data={
+                        'instance_id': new_id,
+                        'instance_create_ts': int(time.time()),
+                    },
+                    completion=False,
+                )
+                constants.instance_id = new_id
         constants.edition = ap.instance_config.data.get('system', {}).get('edition', 'community')
 
         print(f'LangBot instance id: {constants.instance_id}')

src/langbot/pkg/entity/persistence/monitoring.py

@@ -20,6 +20,7 @@ class MonitoringMessage(Base):
     level = sqlalchemy.Column(sqlalchemy.String(50), nullable=False)  # info, warning, error, debug
     platform = sqlalchemy.Column(sqlalchemy.String(255), nullable=True)
     user_id = sqlalchemy.Column(sqlalchemy.String(255), nullable=True)
+    user_name = sqlalchemy.Column(sqlalchemy.String(255), nullable=True)  # User display name
     runner_name = sqlalchemy.Column(sqlalchemy.String(255), nullable=True)  # Runner name for this query
     variables = sqlalchemy.Column(sqlalchemy.Text, nullable=True)  # Query variables as JSON string
     role = sqlalchemy.Column(sqlalchemy.String(50), nullable=True, default='user')  # user, assistant
@@ -64,6 +65,7 @@ class MonitoringSession(Base):
     is_active = sqlalchemy.Column(sqlalchemy.Boolean, nullable=False, default=True, index=True)
     platform = sqlalchemy.Column(sqlalchemy.String(255), nullable=True)
     user_id = sqlalchemy.Column(sqlalchemy.String(255), nullable=True)
+    user_name = sqlalchemy.Column(sqlalchemy.String(255), nullable=True)  # User display name
 
 
 class MonitoringError(Base):

src/langbot/pkg/entity/persistence/rag.py

@@ -10,8 +10,21 @@ class KnowledgeBase(Base):
     emoji = sqlalchemy.Column(sqlalchemy.String(10), nullable=True, default='📚')
     created_at = sqlalchemy.Column(sqlalchemy.DateTime, default=sqlalchemy.func.now())
     updated_at = sqlalchemy.Column(sqlalchemy.DateTime, default=sqlalchemy.func.now(), onupdate=sqlalchemy.func.now())
-    embedding_model_uuid = sqlalchemy.Column(sqlalchemy.String, default='')
-    top_k = sqlalchemy.Column(sqlalchemy.Integer, default=5)
+    # New fields for plugin-based RAG
+    knowledge_engine_plugin_id = sqlalchemy.Column(sqlalchemy.String, nullable=True)
+    collection_id = sqlalchemy.Column(sqlalchemy.String, nullable=True)
+    creation_settings = sqlalchemy.Column(sqlalchemy.JSON, nullable=True, default=None)
+    retrieval_settings = sqlalchemy.Column(sqlalchemy.JSON, nullable=True, default=None)
+
+    # Field sets for different operations
+    MUTABLE_FIELDS = {'name', 'description', 'retrieval_settings'}
+    """Fields that can be updated after creation."""
+
+    CREATE_FIELDS = MUTABLE_FIELDS | {'uuid', 'knowledge_engine_plugin_id', 'collection_id', 'creation_settings'}
+    """Fields used when creating a new knowledge base."""
+
+    ALL_DB_FIELDS = CREATE_FIELDS | {'emoji', 'created_at', 'updated_at'}
+    """All fields stored in database (for loading from DB row)."""
 
 
 class File(Base):
@@ -29,16 +42,3 @@ class Chunk(Base):
     uuid = sqlalchemy.Column(sqlalchemy.String(255), primary_key=True, unique=True)
     file_id = sqlalchemy.Column(sqlalchemy.String(255), nullable=True)
     text = sqlalchemy.Column(sqlalchemy.Text)
-
-
-class ExternalKnowledgeBase(Base):
-    __tablename__ = 'external_knowledge_bases'
-    uuid = sqlalchemy.Column(sqlalchemy.String(255), primary_key=True, unique=True)
-    name = sqlalchemy.Column(sqlalchemy.String, index=True)
-    description = sqlalchemy.Column(sqlalchemy.Text)
-    emoji = sqlalchemy.Column(sqlalchemy.String(10), nullable=True, default='🔗')
-    plugin_author = sqlalchemy.Column(sqlalchemy.String, nullable=False)
-    plugin_name = sqlalchemy.Column(sqlalchemy.String, nullable=False)
-    retriever_name = sqlalchemy.Column(sqlalchemy.String, nullable=False)
-    retriever_config = sqlalchemy.Column(sqlalchemy.JSON, nullable=False, default={})
-    created_at = sqlalchemy.Column(sqlalchemy.DateTime, default=sqlalchemy.func.now())

src/langbot/pkg/persistence/migrations/dbm020_knowledge_engine_plugin_architecture.py

@@ -0,0 +1,161 @@
+import sqlalchemy
+from .. import migration
+
+
+@migration.migration_class(20)
+class DBMigrateKnowledgeEnginePluginArchitecture(migration.DBMigration):
+    """Migrate to unified Knowledge Engine plugin architecture.
+
+    Changes:
+    - Backup existing knowledge_bases data to knowledge_bases_backup
+    - Clear knowledge_bases table and add new plugin architecture columns
+    - Drop old columns (PostgreSQL only; SQLite leaves them unmapped)
+    - Preserve external_knowledge_bases table as-is for future migration
+    - Set rag_plugin_migration_needed flag in metadata if old data exists
+    """
+
+    async def upgrade(self):
+        """Upgrade"""
+        has_internal_data = await self._backup_knowledge_bases()
+        has_external_data = await self._check_external_knowledge_bases()
+        await self._clear_knowledge_bases()
+        await self._add_columns_to_knowledge_bases()
+        await self._drop_old_columns()
+        if has_internal_data or has_external_data:
+            await self._set_migration_flag()
+
+    async def _get_table_columns(self, table_name: str) -> list[str]:
+        """Get column names from a table (works for both SQLite and PostgreSQL)."""
+        if self.ap.persistence_mgr.db.name == 'postgresql':
+            result = await self.ap.persistence_mgr.execute_async(
+                sqlalchemy.text(
+                    'SELECT column_name FROM information_schema.columns WHERE table_name = :table_name;'
+                ).bindparams(table_name=table_name)
+            )
+            return [row[0] for row in result.fetchall()]
+        else:
+            # SQLite PRAGMA does not support bind parameters; validate identifier.
+            if not table_name.isidentifier():
+                raise ValueError(f'Invalid table name: {table_name}')
+            result = await self.ap.persistence_mgr.execute_async(sqlalchemy.text(f'PRAGMA table_info({table_name});'))
+            return [row[1] for row in result.fetchall()]
+
+    async def _table_exists(self, table_name: str) -> bool:
+        """Check if a table exists."""
+        if self.ap.persistence_mgr.db.name == 'postgresql':
+            result = await self.ap.persistence_mgr.execute_async(
+                sqlalchemy.text(
+                    'SELECT EXISTS (SELECT FROM information_schema.tables WHERE table_name = :table_name);'
+                ).bindparams(table_name=table_name)
+            )
+            return result.scalar()
+        else:
+            result = await self.ap.persistence_mgr.execute_async(
+                sqlalchemy.text("SELECT name FROM sqlite_master WHERE type='table' AND name=:table_name;").bindparams(
+                    table_name=table_name
+                )
+            )
+            return result.first() is not None
+
+    async def _backup_knowledge_bases(self) -> bool:
+        """Backup knowledge_bases data. Returns True if data was backed up."""
+        result = await self.ap.persistence_mgr.execute_async(sqlalchemy.text('SELECT COUNT(*) FROM knowledge_bases;'))
+        count = result.scalar()
+        if count == 0:
+            return False
+
+        # Drop backup table if it already exists (from a previous failed migration)
+        if await self._table_exists('knowledge_bases_backup'):
+            await self.ap.persistence_mgr.execute_async(sqlalchemy.text('DROP TABLE knowledge_bases_backup;'))
+
+        await self.ap.persistence_mgr.execute_async(
+            sqlalchemy.text('CREATE TABLE knowledge_bases_backup AS SELECT * FROM knowledge_bases;')
+        )
+        self.ap.logger.info(
+            'Backed up %d knowledge base(s) to knowledge_bases_backup table.',
+            count,
+        )
+        return True
+
+    async def _check_external_knowledge_bases(self) -> bool:
+        """Check if external_knowledge_bases table exists and has data.
+
+        The table is preserved as-is (not dropped) for future migration.
+        """
+        if not await self._table_exists('external_knowledge_bases'):
+            return False
+
+        result = await self.ap.persistence_mgr.execute_async(
+            sqlalchemy.text('SELECT COUNT(*) FROM external_knowledge_bases;')
+        )
+        count = result.scalar()
+        if count > 0:
+            self.ap.logger.info(
+                'Found %d external knowledge base(s) in external_knowledge_bases table. '
+                'Table preserved for future migration.',
+                count,
+            )
+        return count > 0
+
+    async def _clear_knowledge_bases(self):
+        """Clear all rows from knowledge_bases table (preserve table structure)."""
+        await self.ap.persistence_mgr.execute_async(sqlalchemy.text('DELETE FROM knowledge_bases;'))
+
+    async def _add_columns_to_knowledge_bases(self):
+        """Add new RAG plugin architecture columns to knowledge_bases table."""
+        columns = await self._get_table_columns('knowledge_bases')
+
+        new_columns = {
+            'knowledge_engine_plugin_id': 'VARCHAR',
+            'collection_id': 'VARCHAR',
+            'creation_settings': 'TEXT',  # JSON stored as TEXT for SQLite compatibility
+            'retrieval_settings': 'TEXT',
+        }
+
+        for col_name, col_type in new_columns.items():
+            if col_name not in columns:
+                await self.ap.persistence_mgr.execute_async(
+                    sqlalchemy.text(f'ALTER TABLE knowledge_bases ADD COLUMN {col_name} {col_type};')
+                )
+
+    async def _drop_old_columns(self):
+        """Drop embedding_model_uuid and top_k columns (PostgreSQL only).
+
+        SQLite does not support DROP COLUMN in older versions, so we leave the
+        columns in place — the SQLAlchemy entity simply won't map them.
+        """
+        if self.ap.persistence_mgr.db.name != 'postgresql':
+            return
+
+        columns = await self._get_table_columns('knowledge_bases')
+
+        if 'embedding_model_uuid' in columns:
+            await self.ap.persistence_mgr.execute_async(
+                sqlalchemy.text('ALTER TABLE knowledge_bases DROP COLUMN embedding_model_uuid;')
+            )
+
+        if 'top_k' in columns:
+            await self.ap.persistence_mgr.execute_async(
+                sqlalchemy.text('ALTER TABLE knowledge_bases DROP COLUMN top_k;')
+            )
+
+    async def _set_migration_flag(self):
+        """Set rag_plugin_migration_needed flag in metadata table."""
+        # Check if the key already exists
+        result = await self.ap.persistence_mgr.execute_async(
+            sqlalchemy.text("SELECT value FROM metadata WHERE key = 'rag_plugin_migration_needed';")
+        )
+        row = result.first()
+        if row is not None:
+            await self.ap.persistence_mgr.execute_async(
+                sqlalchemy.text("UPDATE metadata SET value = 'true' WHERE key = 'rag_plugin_migration_needed';")
+            )
+        else:
+            await self.ap.persistence_mgr.execute_async(
+                sqlalchemy.text("INSERT INTO metadata (key, value) VALUES ('rag_plugin_migration_needed', 'true');")
+            )
+        self.ap.logger.info('Set rag_plugin_migration_needed=true in metadata.')
+
+    async def downgrade(self):
+        """Downgrade"""
+        pass

src/langbot/pkg/persistence/migrations/dbm021_merge_exception_handling.py

@@ -0,0 +1,74 @@
+from .. import migration
+
+import sqlalchemy
+import json
+
+
+@migration.migration_class(21)
+class DBMigrateMergeExceptionHandling(migration.DBMigration):
+    """Merge hide-exception and block-failed-request-output into a single exception-handling select option,
+    and add failure-hint field.
+
+    Conversion logic:
+    - block-failed-request-output=true  ->  exception-handling: hide
+    - hide-exception=true               ->  exception-handling: show-hint
+    - hide-exception=false              ->  exception-handling: show-error
+    """
+
+    async def upgrade(self):
+        """Upgrade"""
+        result = await self.ap.persistence_mgr.execute_async(
+            sqlalchemy.text('SELECT uuid, config FROM legacy_pipelines')
+        )
+        pipelines = result.fetchall()
+
+        current_version = self.ap.ver_mgr.get_current_version()
+
+        for pipeline_row in pipelines:
+            uuid = pipeline_row[0]
+            config = json.loads(pipeline_row[1]) if isinstance(pipeline_row[1], str) else pipeline_row[1]
+
+            if 'output' not in config:
+                config['output'] = {}
+            if 'misc' not in config['output']:
+                config['output']['misc'] = {}
+
+            misc = config['output']['misc']
+
+            # Determine new exception-handling value from legacy fields
+            hide_exception = misc.get('hide-exception', True)
+            block_failed = misc.get('block-failed-request-output', False)
+
+            if block_failed:
+                exception_handling = 'hide'
+            elif hide_exception:
+                exception_handling = 'show-hint'
+            else:
+                exception_handling = 'show-error'
+
+            misc['exception-handling'] = exception_handling
+
+            # Add failure-hint with default value
+            misc['failure-hint'] = 'Request failed.'
+
+            # Remove legacy fields
+            misc.pop('hide-exception', None)
+
+            if self.ap.persistence_mgr.db.name == 'postgresql':
+                await self.ap.persistence_mgr.execute_async(
+                    sqlalchemy.text(
+                        'UPDATE legacy_pipelines SET config = :config::jsonb, for_version = :for_version WHERE uuid = :uuid'
+                    ),
+                    {'config': json.dumps(config), 'for_version': current_version, 'uuid': uuid},
+                )
+            else:
+                await self.ap.persistence_mgr.execute_async(
+                    sqlalchemy.text(
+                        'UPDATE legacy_pipelines SET config = :config, for_version = :for_version WHERE uuid = :uuid'
+                    ),
+                    {'config': json.dumps(config), 'for_version': current_version, 'uuid': uuid},
+                )
+
+    async def downgrade(self):
+        """Downgrade"""
+        pass

src/langbot/pkg/persistence/migrations/dbm022_monitoring_user_name.py

@@ -0,0 +1,73 @@
+import sqlalchemy
+from .. import migration
+
+
+@migration.migration_class(22)
+class DBMigrateMonitoringUserId(migration.DBMigration):
+    """Add user_id and user_name columns to monitoring_sessions table
+
+    This migration adds the missing user_id column and also ensures user_name
+    column exists (in case migration 21 failed or was skipped).
+    """
+
+    async def _table_exists(self, table_name: str) -> bool:
+        """Check if a table exists (works for both SQLite and PostgreSQL)."""
+        if self.ap.persistence_mgr.db.name == 'postgresql':
+            result = await self.ap.persistence_mgr.execute_async(
+                sqlalchemy.text(
+                    'SELECT EXISTS (SELECT FROM information_schema.tables WHERE table_name = :table_name);'
+                ).bindparams(table_name=table_name)
+            )
+            return bool(result.scalar())
+        else:
+            result = await self.ap.persistence_mgr.execute_async(
+                sqlalchemy.text("SELECT name FROM sqlite_master WHERE type='table' AND name=:table_name;").bindparams(
+                    table_name=table_name
+                )
+            )
+            return result.first() is not None
+
+    async def _get_table_columns(self, table_name: str) -> list[str]:
+        """Get column names from a table (works for both SQLite and PostgreSQL)."""
+        if self.ap.persistence_mgr.db.name == 'postgresql':
+            result = await self.ap.persistence_mgr.execute_async(
+                sqlalchemy.text(
+                    'SELECT column_name FROM information_schema.columns WHERE table_name = :table_name;'
+                ).bindparams(table_name=table_name)
+            )
+            return [row[0] for row in result.fetchall()]
+        else:
+            if not table_name.isidentifier():
+                raise ValueError(f'Invalid table name: {table_name}')
+            result = await self.ap.persistence_mgr.execute_async(sqlalchemy.text(f'PRAGMA table_info({table_name});'))
+            return [row[1] for row in result.fetchall()]
+
+    async def _add_column_if_not_exists(self, table_name: str, column_name: str, column_type: str):
+        """Add a column to a table if it does not already exist."""
+        columns = await self._get_table_columns(table_name)
+        if column_name in columns:
+            self.ap.logger.debug('%s column already exists in %s.', column_name, table_name)
+            return
+        await self.ap.persistence_mgr.execute_async(
+            sqlalchemy.text(f'ALTER TABLE {table_name} ADD COLUMN {column_name} {column_type};')
+        )
+        self.ap.logger.info('Added %s column to %s table.', column_name, table_name)
+
+    async def upgrade(self):
+        # Check if monitoring_sessions table exists
+        if not await self._table_exists('monitoring_sessions'):
+            self.ap.logger.warning('monitoring_sessions table does not exist, skipping migration.')
+            return
+
+        # Add user_id column to monitoring_sessions table
+        await self._add_column_if_not_exists('monitoring_sessions', 'user_id', 'VARCHAR(255)')
+
+        # Add user_name column to monitoring_sessions table (in case migration 21 failed)
+        await self._add_column_if_not_exists('monitoring_sessions', 'user_name', 'VARCHAR(255)')
+
+        # Add user_name column to monitoring_messages table (in case migration 21 failed)
+        if await self._table_exists('monitoring_messages'):
+            await self._add_column_if_not_exists('monitoring_messages', 'user_name', 'VARCHAR(255)')
+
+    async def downgrade(self):
+        pass

src/langbot/pkg/persistence/migrations/dbm023_model_fallback_config.py

@@ -0,0 +1,102 @@
+from .. import migration
+
+import sqlalchemy
+import json
+
+
+@migration.migration_class(23)
+class DBMigrateModelFallbackConfig(migration.DBMigration):
+    """Convert model field from plain UUID string to object with primary/fallbacks"""
+
+    async def upgrade(self):
+        """Upgrade"""
+        result = await self.ap.persistence_mgr.execute_async(
+            sqlalchemy.text('SELECT uuid, config FROM legacy_pipelines')
+        )
+        pipelines = result.fetchall()
+
+        current_version = self.ap.ver_mgr.get_current_version()
+
+        for pipeline_row in pipelines:
+            uuid = pipeline_row[0]
+            config = json.loads(pipeline_row[1]) if isinstance(pipeline_row[1], str) else pipeline_row[1]
+
+            if 'ai' not in config or 'local-agent' not in config['ai']:
+                continue
+
+            local_agent = config['ai']['local-agent']
+            changed = False
+
+            # Convert model from string to object
+            model_value = local_agent.get('model', '')
+            if isinstance(model_value, str):
+                local_agent['model'] = {
+                    'primary': model_value,
+                    'fallbacks': [],
+                }
+                changed = True
+
+            # Remove leftover fallback-models field if present
+            if 'fallback-models' in local_agent:
+                del local_agent['fallback-models']
+                changed = True
+
+            if not changed:
+                continue
+
+            # Update using raw SQL with compatibility for both SQLite and PostgreSQL
+            if self.ap.persistence_mgr.db.name == 'postgresql':
+                await self.ap.persistence_mgr.execute_async(
+                    sqlalchemy.text(
+                        'UPDATE legacy_pipelines SET config = :config::jsonb, for_version = :for_version WHERE uuid = :uuid'
+                    ),
+                    {'config': json.dumps(config), 'for_version': current_version, 'uuid': uuid},
+                )
+            else:
+                await self.ap.persistence_mgr.execute_async(
+                    sqlalchemy.text(
+                        'UPDATE legacy_pipelines SET config = :config, for_version = :for_version WHERE uuid = :uuid'
+                    ),
+                    {'config': json.dumps(config), 'for_version': current_version, 'uuid': uuid},
+                )
+
+    async def downgrade(self):
+        """Downgrade"""
+        result = await self.ap.persistence_mgr.execute_async(
+            sqlalchemy.text('SELECT uuid, config FROM legacy_pipelines')
+        )
+        pipelines = result.fetchall()
+
+        current_version = self.ap.ver_mgr.get_current_version()
+
+        for pipeline_row in pipelines:
+            uuid = pipeline_row[0]
+            config = json.loads(pipeline_row[1]) if isinstance(pipeline_row[1], str) else pipeline_row[1]
+
+            if 'ai' not in config or 'local-agent' not in config['ai']:
+                continue
+
+            local_agent = config['ai']['local-agent']
+
+            # Convert model from object back to string
+            model_value = local_agent.get('model', '')
+            if isinstance(model_value, dict):
+                local_agent['model'] = model_value.get('primary', '')
+            else:
+                continue
+
+            # Update using raw SQL with compatibility for both SQLite and PostgreSQL
+            if self.ap.persistence_mgr.db.name == 'postgresql':
+                await self.ap.persistence_mgr.execute_async(
+                    sqlalchemy.text(
+                        'UPDATE legacy_pipelines SET config = :config::jsonb, for_version = :for_version WHERE uuid = :uuid'
+                    ),
+                    {'config': json.dumps(config), 'for_version': current_version, 'uuid': uuid},
+                )
+            else:
+                await self.ap.persistence_mgr.execute_async(
+                    sqlalchemy.text(
+                        'UPDATE legacy_pipelines SET config = :config, for_version = :for_version WHERE uuid = :uuid'
+                    ),
+                    {'config': json.dumps(config), 'for_version': current_version, 'uuid': uuid},
+                )

src/langbot/pkg/persistence/migrations/dbm024_wecombot_websocket_mode.py

@@ -0,0 +1,49 @@
+from .. import migration
+
+import sqlalchemy
+import json
+
+
+@migration.migration_class(24)
+class DBMigrateWecomBotWebSocketMode(migration.DBMigration):
+    """Add enable-webhook field to existing wecombot adapter configs.
+
+    Existing wecombot bots were all using webhook mode, so we set
+    enable-webhook=true to preserve their behavior after the new
+    WebSocket long connection mode is introduced as default.
+    """
+
+    async def upgrade(self):
+        """Upgrade"""
+        result = await self.ap.persistence_mgr.execute_async(
+            sqlalchemy.text("SELECT uuid, adapter_config FROM bots WHERE adapter = 'wecombot'")
+        )
+        bots = result.fetchall()
+
+        for bot_row in bots:
+            bot_uuid = bot_row[0]
+            adapter_config = json.loads(bot_row[1]) if isinstance(bot_row[1], str) else bot_row[1]
+
+            if 'enable-webhook' in adapter_config:
+                continue
+
+            # Determine mode based on existing config: if webhook fields are present, keep webhook mode
+            has_webhook_config = bool(
+                adapter_config.get('Token') and adapter_config.get('EncodingAESKey') and adapter_config.get('Corpid')
+            )
+            adapter_config['enable-webhook'] = has_webhook_config
+
+            if self.ap.persistence_mgr.db.name == 'postgresql':
+                await self.ap.persistence_mgr.execute_async(
+                    sqlalchemy.text('UPDATE bots SET adapter_config = :config::jsonb WHERE uuid = :uuid'),
+                    {'config': json.dumps(adapter_config), 'uuid': bot_uuid},
+                )
+            else:
+                await self.ap.persistence_mgr.execute_async(
+                    sqlalchemy.text('UPDATE bots SET adapter_config = :config WHERE uuid = :uuid'),
+                    {'config': json.dumps(adapter_config), 'uuid': bot_uuid},
+                )
+
+    async def downgrade(self):
+        """Downgrade"""
+        pass

src/langbot/pkg/pipeline/cntfilter/filters/baiduexamine.py

@@ -1,10 +1,9 @@
 from __future__ import annotations
 
-import aiohttp
-
 from .. import entities
 from .. import filter as filter_model
 import langbot_plugin.api.entities.builtin.pipeline.query as pipeline_query
+from langbot.pkg.utils import httpclient
 
 BAIDU_EXAMINE_URL = 'https://aip.baidubce.com/rest/2.0/solution/v1/text_censor/v2/user_defined?access_token={}'
 BAIDU_EXAMINE_TOKEN_URL = 'https://aip.baidubce.com/oauth/2.0/token'
@@ -15,50 +14,50 @@ class BaiduCloudExamine(filter_model.ContentFilter):
     """百度云内容审核"""
 
     async def _get_token(self) -> str:
-        async with aiohttp.ClientSession() as session:
-            async with session.post(
-                BAIDU_EXAMINE_TOKEN_URL,
-                params={
-                    'grant_type': 'client_credentials',
-                    'client_id': self.ap.pipeline_cfg.data['baidu-cloud-examine']['api-key'],
-                    'client_secret': self.ap.pipeline_cfg.data['baidu-cloud-examine']['api-secret'],
-                },
-            ) as resp:
-                return (await resp.json())['access_token']
+        session = httpclient.get_session()
+        async with session.post(
+            BAIDU_EXAMINE_TOKEN_URL,
+            params={
+                'grant_type': 'client_credentials',
+                'client_id': self.ap.pipeline_cfg.data['baidu-cloud-examine']['api-key'],
+                'client_secret': self.ap.pipeline_cfg.data['baidu-cloud-examine']['api-secret'],
+            },
+        ) as resp:
+            return (await resp.json())['access_token']
 
     async def process(self, query: pipeline_query.Query, message: str) -> entities.FilterResult:
-        async with aiohttp.ClientSession() as session:
-            async with session.post(
-                BAIDU_EXAMINE_URL.format(await self._get_token()),
-                headers={
-                    'Content-Type': 'application/x-www-form-urlencoded',
-                    'Accept': 'application/json',
-                },
-                data=f'text={message}'.encode('utf-8'),
-            ) as resp:
-                result = await resp.json()
+        session = httpclient.get_session()
+        async with session.post(
+            BAIDU_EXAMINE_URL.format(await self._get_token()),
+            headers={
+                'Content-Type': 'application/x-www-form-urlencoded',
+                'Accept': 'application/json',
+            },
+            data=f'text={message}'.encode('utf-8'),
+        ) as resp:
+            result = await resp.json()
 
-                if 'error_code' in result:
+            if 'error_code' in result:
+                return entities.FilterResult(
+                    level=entities.ResultLevel.BLOCK,
+                    replacement=message,
+                    user_notice='',
+                    console_notice=f'百度云判定出错，错误信息：{result["error_msg"]}',
+                )
+            else:
+                conclusion = result['conclusion']
+
+                if conclusion in ('合规'):
+                    return entities.FilterResult(
+                        level=entities.ResultLevel.PASS,
+                        replacement=message,
+                        user_notice='',
+                        console_notice=f'百度云判定结果：{conclusion}',
+                    )
+                else:
                     return entities.FilterResult(
                         level=entities.ResultLevel.BLOCK,
                         replacement=message,
-                        user_notice='',
-                        console_notice=f'百度云判定出错，错误信息：{result["error_msg"]}',
+                        user_notice='消息中存在不合适的内容, 请修改',
+                        console_notice=f'百度云判定结果：{conclusion}',
                     )
-                else:
-                    conclusion = result['conclusion']
-
-                    if conclusion in ('合规'):
-                        return entities.FilterResult(
-                            level=entities.ResultLevel.PASS,
-                            replacement=message,
-                            user_notice='',
-                            console_notice=f'百度云判定结果：{conclusion}',
-                        )
-                    else:
-                        return entities.FilterResult(
-                            level=entities.ResultLevel.BLOCK,
-                            replacement=message,
-                            user_notice='消息中存在不合适的内容, 请修改',
-                            console_notice=f'百度云判定结果：{conclusion}',
-                        )

src/langbot/pkg/pipeline/config_coercion.py

@@ -0,0 +1,105 @@
+from __future__ import annotations
+
+import logging
+
+logger = logging.getLogger(__name__)
+
+# metadata type -> coercion function
+_COERCE_MAP = {
+    'integer': lambda v: int(v),
+    'number': lambda v: float(v),
+    'float': lambda v: float(v),
+}
+
+
+def _coerce_bool(v):
+    if isinstance(v, bool):
+        return v
+    if isinstance(v, str):
+        if v.lower() == 'true':
+            return True
+        if v.lower() == 'false':
+            return False
+        raise ValueError(f'Cannot convert string {v!r} to bool')
+    return bool(v)
+
+
+def _coerce_value(value, expected_type: str):
+    """Convert a single value to the expected type.
+
+    Returns the converted value, or the original value if no conversion needed.
+    """
+    if value is None:
+        return value
+
+    if expected_type == 'boolean':
+        if isinstance(value, bool):
+            return value
+        return _coerce_bool(value)
+
+    coerce_fn = _COERCE_MAP.get(expected_type)
+    if coerce_fn is None:
+        return value
+
+    # Already the correct type
+    if expected_type == 'integer' and isinstance(value, int) and not isinstance(value, bool):
+        return value
+    if expected_type in ('number', 'float') and isinstance(value, (int, float)) and not isinstance(value, bool):
+        return float(value)
+
+    return coerce_fn(value)
+
+
+def coerce_pipeline_config(
+    config: dict,
+    *metadata_list: dict,
+) -> None:
+    """Coerce pipeline config values according to metadata type definitions.
+
+    Walks each metadata dict (trigger, safety, ai, output) and converts
+    config values in-place so that strings coming from the JSON column are
+    cast to their declared types (integer, number/float, boolean).
+
+    Args:
+        config: The pipeline config dict to modify in-place.
+        *metadata_list: Metadata dicts loaded from the YAML templates.
+    """
+    for meta in metadata_list:
+        section_name = meta.get('name')
+        if not section_name or section_name not in config:
+            continue
+
+        section = config[section_name]
+        if not isinstance(section, dict):
+            continue
+
+        for stage_def in meta.get('stages', []):
+            stage_name = stage_def.get('name')
+            if not stage_name or stage_name not in section:
+                continue
+
+            stage_config = section[stage_name]
+            if not isinstance(stage_config, dict):
+                continue
+
+            for field_def in stage_def.get('config', []):
+                field_name = field_def.get('name')
+                field_type = field_def.get('type')
+                if not field_name or not field_type or field_name not in stage_config:
+                    continue
+
+                old_value = stage_config[field_name]
+                try:
+                    new_value = _coerce_value(old_value, field_type)
+                    if new_value is not old_value:
+                        stage_config[field_name] = new_value
+                except (ValueError, TypeError) as e:
+                    logger.warning(
+                        'Failed to coerce config %s.%s.%s (%r) to %s: %s',
+                        section_name,
+                        stage_name,
+                        field_name,
+                        old_value,
+                        field_type,
+                        e,
+                    )

src/langbot/pkg/pipeline/monitoring_helper.py

@@ -34,6 +34,15 @@ class MonitoringHelper:
             # Check if session exists, if not, record session start
             session_id = f'{query.launcher_type}_{query.launcher_id}'
 
+            # Get sender name from message event
+            sender_name = None
+            if hasattr(query, 'message_event'):
+                if hasattr(query.message_event, 'sender'):
+                    if hasattr(query.message_event.sender, 'nickname'):
+                        sender_name = query.message_event.sender.nickname
+                    elif hasattr(query.message_event.sender, 'member_name'):
+                        sender_name = query.message_event.sender.member_name
+
             # Try to record message
             # Use JSON serialization to preserve message chain structure (including image URLs, etc.)
             if hasattr(query, 'message_chain') and hasattr(query.message_chain, 'model_dump'):
@@ -57,6 +66,7 @@ class MonitoringHelper:
                 if hasattr(query.launcher_type, 'value')
                 else str(query.launcher_type),
                 user_id=query.sender_id,
+                user_name=sender_name,
                 runner_name=runner_name,
                 variables=None,  # Will be updated in record_query_success
             )
@@ -80,6 +90,7 @@ class MonitoringHelper:
                     if hasattr(query.launcher_type, 'value')
                     else str(query.launcher_type),
                     user_id=query.sender_id,
+                    user_name=sender_name,
                 )
 
             return message_id
@@ -128,6 +139,15 @@ class MonitoringHelper:
         try:
             session_id = f'{query.launcher_type}_{query.launcher_id}'
 
+            # Get sender name from message event
+            sender_name = None
+            if hasattr(query, 'message_event'):
+                if hasattr(query.message_event, 'sender'):
+                    if hasattr(query.message_event.sender, 'nickname'):
+                        sender_name = query.message_event.sender.nickname
+                    elif hasattr(query.message_event.sender, 'member_name'):
+                        sender_name = query.message_event.sender.member_name
+
             # Extract response content from resp_message_chain
             if hasattr(query, 'resp_message_chain') and query.resp_message_chain:
                 # Serialize the last response message chain
@@ -162,6 +182,7 @@ class MonitoringHelper:
                 if hasattr(query.launcher_type, 'value')
                 else str(query.launcher_type),
                 user_id=query.sender_id,
+                user_name=sender_name,
                 runner_name=runner_name,
                 role='assistant',
             )
@@ -183,6 +204,15 @@ class MonitoringHelper:
         try:
             session_id = f'{query.launcher_type}_{query.launcher_id}'
 
+            # Get sender name from message event
+            sender_name = None
+            if hasattr(query, 'message_event'):
+                if hasattr(query.message_event, 'sender'):
+                    if hasattr(query.message_event.sender, 'nickname'):
+                        sender_name = query.message_event.sender.nickname
+                    elif hasattr(query.message_event.sender, 'member_name'):
+                        sender_name = query.message_event.sender.member_name
+
             # Record error message
             message_id = await ap.monitoring_service.record_message(
                 bot_id=bot_id,
@@ -197,6 +227,7 @@ class MonitoringHelper:
                 if hasattr(query.launcher_type, 'value')
                 else str(query.launcher_type),
                 user_id=query.sender_id,
+                user_name=sender_name,
                 runner_name=runner_name,
             )
 

src/langbot/pkg/pipeline/pipelinemgr.py

@@ -13,6 +13,7 @@ import langbot_plugin.api.entities.builtin.platform.message as platform_message
 import langbot_plugin.api.entities.builtin.platform.events as platform_events
 import langbot_plugin.api.entities.events as events
 from ..utils import importutil
+from .config_coercion import coerce_pipeline_config
 
 import langbot_plugin.api.entities.builtin.provider.session as provider_session
 import langbot_plugin.api.entities.builtin.pipeline.query as pipeline_query
@@ -420,6 +421,14 @@ class PipelineManager:
         elif isinstance(pipeline_entity, dict):
             pipeline_entity = persistence_pipeline.LegacyPipeline(**pipeline_entity)
 
+        coerce_pipeline_config(
+            pipeline_entity.config,
+            getattr(self.ap, 'pipeline_config_meta_trigger', {'name': 'trigger', 'stages': []}),
+            getattr(self.ap, 'pipeline_config_meta_safety', {'name': 'safety', 'stages': []}),
+            getattr(self.ap, 'pipeline_config_meta_ai', {'name': 'ai', 'stages': []}),
+            getattr(self.ap, 'pipeline_config_meta_output', {'name': 'output', 'stages': []}),
+        )
+
         # initialize stage containers according to pipeline_entity.stages
         stage_containers: list[StageInstContainer] = []
         for stage_name in pipeline_entity.stages:

src/langbot/pkg/pipeline/preproc/preproc.py

@@ -36,17 +36,36 @@ class PreProcessor(stage.PipelineStage):
         session = await self.ap.sess_mgr.get_session(query)
 
         # When not local-agent, llm_model is None
-        try:
-            llm_model = (
-                await self.ap.model_mgr.get_model_by_uuid(query.pipeline_config['ai']['local-agent']['model'])
-                if selected_runner == 'local-agent'
-                else None
-            )
-        except ValueError:
-            self.ap.logger.warning(
-                f'LLM model {query.pipeline_config["ai"]["local-agent"]["model"] + " "}not found or not configured'
-            )
-            llm_model = None
+        llm_model = None
+        if selected_runner == 'local-agent':
+            # Read model config — new format is { primary: str, fallbacks: [str] },
+            # but handle legacy plain string for backward compatibility
+            model_config = query.pipeline_config['ai']['local-agent'].get('model', {})
+            if isinstance(model_config, str):
+                # Legacy format: plain UUID string
+                primary_uuid = model_config
+                fallback_uuids = []
+            else:
+                primary_uuid = model_config.get('primary', '')
+                fallback_uuids = model_config.get('fallbacks', [])
+
+            if primary_uuid:
+                try:
+                    llm_model = await self.ap.model_mgr.get_model_by_uuid(primary_uuid)
+                except ValueError:
+                    self.ap.logger.warning(f'LLM model {primary_uuid} not found or not configured')
+
+            # Resolve fallback model UUIDs
+            if fallback_uuids:
+                valid_fallbacks = []
+                for fb_uuid in fallback_uuids:
+                    try:
+                        await self.ap.model_mgr.get_model_by_uuid(fb_uuid)
+                        valid_fallbacks.append(fb_uuid)
+                    except ValueError:
+                        self.ap.logger.warning(f'Fallback model {fb_uuid} not found, skipping')
+                if valid_fallbacks:
+                    query.variables['_fallback_model_uuids'] = valid_fallbacks
 
         conversation = await self.ap.sess_mgr.get_conversation(
             query,
@@ -61,20 +80,28 @@ class PreProcessor(stage.PipelineStage):
         query.prompt = conversation.prompt.copy()
         query.messages = conversation.messages.copy()
 
-        if selected_runner == 'local-agent' and llm_model:
+        if selected_runner == 'local-agent':
             query.use_funcs = []
-            query.use_llm_model_uuid = llm_model.model_entity.uuid
+            if llm_model:
+                query.use_llm_model_uuid = llm_model.model_entity.uuid
 
-            if llm_model.model_entity.abilities.__contains__('func_call'):
-                # Get bound plugins and MCP servers for filtering tools
+                if llm_model.model_entity.abilities.__contains__('func_call'):
+                    # Get bound plugins and MCP servers for filtering tools
+                    bound_plugins = query.variables.get('_pipeline_bound_plugins', None)
+                    bound_mcp_servers = query.variables.get('_pipeline_bound_mcp_servers', None)
+                    query.use_funcs = await self.ap.tool_mgr.get_all_tools(bound_plugins, bound_mcp_servers)
+
+                    self.ap.logger.debug(f'Bound plugins: {bound_plugins}')
+                    self.ap.logger.debug(f'Bound MCP servers: {bound_mcp_servers}')
+                    self.ap.logger.debug(f'Use funcs: {query.use_funcs}')
+
+            # If primary model doesn't support func_call but fallback models exist,
+            # load tools anyway since fallback models may support them
+            if not query.use_funcs and query.variables.get('_fallback_model_uuids'):
                 bound_plugins = query.variables.get('_pipeline_bound_plugins', None)
                 bound_mcp_servers = query.variables.get('_pipeline_bound_mcp_servers', None)
                 query.use_funcs = await self.ap.tool_mgr.get_all_tools(bound_plugins, bound_mcp_servers)
 
-                self.ap.logger.debug(f'Bound plugins: {bound_plugins}')
-                self.ap.logger.debug(f'Bound MCP servers: {bound_mcp_servers}')
-                self.ap.logger.debug(f'Use funcs: {query.use_funcs}')
-
         sender_name = ''
 
         if isinstance(query.message_event, platform_events.GroupMessage):
@@ -149,6 +176,16 @@ class PreProcessor(stage.PipelineStage):
         query.variables['user_message_text'] = plain_text
 
         query.user_message = provider_message.Message(role='user', content=content_list)
+
+        # Extract knowledge base UUIDs into query variables so plugins can modify them
+        # during PromptPreProcessing before the runner performs retrieval.
+        kb_uuids = query.pipeline_config['ai']['local-agent'].get('knowledge-bases', [])
+        if not kb_uuids:
+            old_kb_uuid = query.pipeline_config['ai']['local-agent'].get('knowledge-base', '')
+            if old_kb_uuid and old_kb_uuid != '__none__':
+                kb_uuids = [old_kb_uuid]
+        query.variables['_knowledge_base_uuids'] = list(kb_uuids)
+
         # =========== 触发事件 PromptPreProcessing
 
         event = events.PromptPreProcessing(

src/langbot/pkg/pipeline/process/handlers/chat.py

@@ -12,7 +12,7 @@ from ... import entities
 from ....provider import runner as runner_module
 
 import langbot_plugin.api.entities.events as events
-from ....utils import importutil, constants
+from ....utils import importutil, constants, runner as runner_utils
 from ....provider import runners
 import langbot_plugin.api.entities.builtin.provider.session as provider_session
 import langbot_plugin.api.entities.builtin.pipeline.query as pipeline_query
@@ -149,12 +149,19 @@ class ChatMessageHandler(handler.MessageHandler):
                 self.ap.logger.error(f'Conversation({query.query_id}) Request Failed: {error_info}')
                 traceback.print_exc()
 
-                hide_exception_info = query.pipeline_config['output']['misc']['hide-exception']
+                exception_handling = query.pipeline_config['output']['misc'].get('exception-handling', 'show-hint')
+
+                if exception_handling == 'show-error':
+                    user_notice = f'{e}'
+                elif exception_handling == 'show-hint':
+                    user_notice = query.pipeline_config['output']['misc'].get('failure-hint', 'Request failed.')
+                else:  # hide
+                    user_notice = None
 
                 yield entities.StageProcessResult(
                     result_type=entities.ResultType.INTERRUPT,
                     new_query=query,
-                    user_notice='请求失败' if hide_exception_info else f'{e}',
+                    user_notice=user_notice,
                     error_notice=f'{e}',
                     debug_notice=traceback.format_exc(),
                 )
@@ -185,10 +192,15 @@ class ChatMessageHandler(handler.MessageHandler):
 
                     pipeline_plugins = query.variables.get('_pipeline_bound_plugins', None)
 
+                    runner_category = runner_utils.get_runner_category_from_runner(
+                        runner_name, runner, query.pipeline_config
+                    )
+
                     payload = {
                         'query_id': query.query_id,
                         'adapter': adapter_name,
                         'runner': runner_name,
+                        'runner_category': runner_category,
                         'duration_ms': duration_ms,
                         'model_name': model_name,
                         'version': constants.semantic_version,

src/langbot/pkg/platform/botmgr.py

@@ -282,6 +282,8 @@ class PlatformManager:
         return runtime_bot
 
     async def get_bot_by_uuid(self, bot_uuid: str) -> RuntimeBot | None:
+        if self.websocket_proxy_bot and self.websocket_proxy_bot.bot_entity.uuid == bot_uuid:
+            return self.websocket_proxy_bot
         for bot in self.bots:
             if bot.bot_entity.uuid == bot_uuid:
                 return bot

src/langbot/pkg/platform/sources/discord.py

@@ -14,7 +14,7 @@ import io
 import asyncio
 from enum import Enum
 
-import aiohttp
+from langbot.pkg.utils import httpclient
 import pydantic
 
 import langbot_plugin.api.definition.abstract.platform.adapter as abstract_platform_adapter
@@ -622,23 +622,23 @@ class DiscordMessageConverter(abstract_platform_adapter.AbstractMessageConverter
                     image_bytes = base64.b64decode(base64_data)
                 elif ele.url:
                     # 从URL下载图片
-                    async with aiohttp.ClientSession() as session:
-                        async with session.get(ele.url) as response:
-                            image_bytes = await response.read()
-                            # 从URL或Content-Type推断文件类型
-                            content_type = response.headers.get('Content-Type', '')
-                            if 'jpeg' in content_type or 'jpg' in content_type:
-                                filename = f'{uuid.uuid4()}.jpg'
-                            elif 'gif' in content_type:
-                                filename = f'{uuid.uuid4()}.gif'
-                            elif 'webp' in content_type:
-                                filename = f'{uuid.uuid4()}.webp'
-                            elif ele.url.lower().endswith(('.jpg', '.jpeg')):
-                                filename = f'{uuid.uuid4()}.jpg'
-                            elif ele.url.lower().endswith('.gif'):
-                                filename = f'{uuid.uuid4()}.gif'
-                            elif ele.url.lower().endswith('.webp'):
-                                filename = f'{uuid.uuid4()}.webp'
+                    session = httpclient.get_session()
+                    async with session.get(ele.url) as response:
+                        image_bytes = await response.read()
+                        # 从URL或Content-Type推断文件类型
+                        content_type = response.headers.get('Content-Type', '')
+                        if 'jpeg' in content_type or 'jpg' in content_type:
+                            filename = f'{uuid.uuid4()}.jpg'
+                        elif 'gif' in content_type:
+                            filename = f'{uuid.uuid4()}.gif'
+                        elif 'webp' in content_type:
+                            filename = f'{uuid.uuid4()}.webp'
+                        elif ele.url.lower().endswith(('.jpg', '.jpeg')):
+                            filename = f'{uuid.uuid4()}.jpg'
+                        elif ele.url.lower().endswith('.gif'):
+                            filename = f'{uuid.uuid4()}.gif'
+                        elif ele.url.lower().endswith('.webp'):
+                            filename = f'{uuid.uuid4()}.webp'
                 elif ele.path:
                     # 从文件路径读取图片
                     # 确保路径没有空字节
@@ -702,9 +702,9 @@ class DiscordMessageConverter(abstract_platform_adapter.AbstractMessageConverter
                     file_base64 = ele.base64.split(',')[-1]
                     file_bytes = base64.b64decode(file_base64)
                 elif ele.url:
-                    async with aiohttp.ClientSession() as session:
-                        async with session.get(ele.url) as response:
-                            file_bytes = await response.read()
+                    session = httpclient.get_session()
+                    async with session.get(ele.url) as response:
+                        file_bytes = await response.read()
                 if file_bytes:
                     files.append(discord.File(fp=io.BytesIO(file_bytes), filename=filename))
             elif isinstance(ele, platform_message.File):
@@ -717,9 +717,9 @@ class DiscordMessageConverter(abstract_platform_adapter.AbstractMessageConverter
                     else:
                         file_bytes = base64.b64decode(ele.base64)
                 elif ele.url:
-                    async with aiohttp.ClientSession() as session:
-                        async with session.get(ele.url) as response:
-                            file_bytes = await response.read()
+                    session = httpclient.get_session()
+                    async with session.get(ele.url) as response:
+                        file_bytes = await response.read()
                 if file_bytes:
                     files.append(discord.File(fp=io.BytesIO(file_bytes), filename=filename))
             elif isinstance(ele, platform_message.Forward):
@@ -775,12 +775,12 @@ class DiscordMessageConverter(abstract_platform_adapter.AbstractMessageConverter
 
         # attachments
         for attachment in message.attachments:
-            async with aiohttp.ClientSession(trust_env=True) as session:
-                async with session.get(attachment.url) as response:
-                    image_data = await response.read()
-                    image_base64 = base64.b64encode(image_data).decode('utf-8')
-                    image_format = response.headers['Content-Type']
-                    element_list.append(platform_message.Image(base64=f'data:{image_format};base64,{image_base64}'))
+            session = httpclient.get_session(trust_env=True)
+            async with session.get(attachment.url) as response:
+                image_data = await response.read()
+                image_base64 = base64.b64encode(image_data).decode('utf-8')
+                image_format = response.headers['Content-Type']
+                element_list.append(platform_message.Image(base64=f'data:{image_format};base64,{image_base64}'))
 
         return platform_message.MessageChain(element_list)
 

src/langbot/pkg/platform/sources/kook.py

@@ -9,6 +9,8 @@ import traceback
 import time
 
 import aiohttp
+
+from langbot.pkg.utils import httpclient
 import websockets
 import pydantic
 
@@ -120,16 +122,16 @@ class KookMessageConverter(abstract_platform_adapter.AbstractMessageConverter):
             if content:
                 # Download image and convert to base64
                 try:
-                    async with aiohttp.ClientSession() as session:
-                        async with session.get(content) as response:
-                            if response.status == 200:
-                                image_bytes = await response.read()
-                                image_base64 = base64.b64encode(image_bytes).decode('utf-8')
-                                # Detect image format
-                                content_type = response.headers.get('Content-Type', 'image/png')
-                                components.append(
-                                    platform_message.Image(base64=f'data:{content_type};base64,{image_base64}')
-                                )
+                    session = httpclient.get_session()
+                    async with session.get(content) as response:
+                        if response.status == 200:
+                            image_bytes = await response.read()
+                            image_base64 = base64.b64encode(image_bytes).decode('utf-8')
+                            # Detect image format
+                            content_type = response.headers.get('Content-Type', 'image/png')
+                            components.append(
+                                platform_message.Image(base64=f'data:{content_type};base64,{image_base64}')
+                            )
                 except Exception:
                     # If download fails, just add as plain text
                     components.append(platform_message.Plain(text=f'[Image: {content}]'))
@@ -295,17 +297,17 @@ class KookAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter):
             'Authorization': f'Bot {self.config["token"]}',
         }
 
-        async with aiohttp.ClientSession() as session:
-            async with session.get(base_url, params=params, headers=headers) as response:
-                if response.status == 200:
-                    data = await response.json()
-                    if data.get('code') == 0:
-                        gateway_url = data['data']['url']
-                        return gateway_url
-                    else:
-                        raise Exception(f'Failed to get gateway URL: {data.get("message")}')
+        session = httpclient.get_session()
+        async with session.get(base_url, params=params, headers=headers) as response:
+            if response.status == 200:
+                data = await response.json()
+                if data.get('code') == 0:
+                    gateway_url = data['data']['url']
+                    return gateway_url
                 else:
-                    raise Exception(f'Failed to get gateway URL: HTTP {response.status}')
+                    raise Exception(f'Failed to get gateway URL: {data.get("message")}')
+            else:
+                raise Exception(f'Failed to get gateway URL: HTTP {response.status}')
 
     async def _get_bot_user_info(self) -> dict:
         """Get bot's own user information from KOOK API"""
@@ -315,17 +317,17 @@ class KookAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter):
             'Authorization': f'Bot {self.config["token"]}',
         }
 
-        async with aiohttp.ClientSession() as session:
-            async with session.get(base_url, headers=headers) as response:
-                if response.status == 200:
-                    data = await response.json()
-                    if data.get('code') == 0:
-                        user_info = data['data']
-                        return user_info
-                    else:
-                        raise Exception(f'Failed to get bot user info: {data.get("message")}')
+        session = httpclient.get_session()
+        async with session.get(base_url, headers=headers) as response:
+            if response.status == 200:
+                data = await response.json()
+                if data.get('code') == 0:
+                    user_info = data['data']
+                    return user_info
                 else:
-                    raise Exception(f'Failed to get bot user info: HTTP {response.status}')
+                    raise Exception(f'Failed to get bot user info: {data.get("message")}')
+            else:
+                raise Exception(f'Failed to get bot user info: HTTP {response.status}')
 
     async def _handle_hello(self, data: dict):
         """Handle HELLO signal (signal 1)"""
@@ -510,7 +512,7 @@ class KookAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter):
 
         try:
             if not self.http_session:
-                self.http_session = aiohttp.ClientSession()
+                self.http_session = httpclient.get_session()
 
             async with self.http_session.post(url, json=payload, headers=headers) as response:
                 if response.status == 200:
@@ -576,7 +578,7 @@ class KookAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter):
 
         try:
             if not self.http_session:
-                self.http_session = aiohttp.ClientSession()
+                self.http_session = httpclient.get_session()
 
             async with self.http_session.post(url, json=payload, headers=headers) as response:
                 if response.status == 200:
@@ -624,7 +626,7 @@ class KookAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter):
 
         try:
             # Create HTTP session
-            self.http_session = aiohttp.ClientSession()
+            self.http_session = httpclient.get_session()
 
             await self.logger.info('Starting KOOK adapter')
 

src/langbot/pkg/platform/sources/lark.py

@@ -17,7 +17,7 @@ import tempfile
 import os
 import mimetypes
 
-import aiohttp
+from langbot.pkg.utils import httpclient
 import lark_oapi.ws.exception
 import quart
 from lark_oapi.api.im.v1 import *
@@ -78,13 +78,13 @@ class LarkMessageConverter(abstract_platform_adapter.AbstractMessageConverter):
                 return None
         elif msg.url:
             try:
-                async with aiohttp.ClientSession() as session:
-                    async with session.get(msg.url) as response:
-                        if response.status == 200:
-                            image_bytes = await response.read()
-                        else:
-                            print(f'Failed to download image from {msg.url}: HTTP {response.status}')
-                            return None
+                session = httpclient.get_session()
+                async with session.get(msg.url) as response:
+                    if response.status == 200:
+                        image_bytes = await response.read()
+                    else:
+                        print(f'Failed to download image from {msg.url}: HTTP {response.status}')
+                        return None
             except Exception as e:
                 print(f'Failed to download image from {msg.url}: {e}')
                 traceback.print_exc()
@@ -208,10 +208,10 @@ class LarkMessageConverter(abstract_platform_adapter.AbstractMessageConverter):
                 pass
         elif msg.url:
             try:
-                async with aiohttp.ClientSession() as session:
-                    async with session.get(msg.url) as resp:
-                        if resp.status == 200:
-                            data = await resp.read()
+                session = httpclient.get_session()
+                async with session.get(msg.url) as resp:
+                    if resp.status == 200:
+                        data = await resp.read()
             except Exception:
                 pass
         elif msg.path:
@@ -575,6 +575,127 @@ class LarkMessageConverter(abstract_platform_adapter.AbstractMessageConverter):
 
 
 class LarkEventConverter(abstract_platform_adapter.AbstractEventConverter):
+    _processed_thread_quote_cache: typing.ClassVar[dict[str, float]] = {}
+    _processed_thread_quote_cache_max_size: typing.ClassVar[int] = 4096
+    _processed_thread_quote_cache_ttl_seconds: typing.ClassVar[int] = 86400
+
+    @classmethod
+    def _prune_processed_thread_quote_cache(cls, now: typing.Optional[float] = None) -> None:
+        if now is None:
+            now = time.time()
+
+        expire_before = now - cls._processed_thread_quote_cache_ttl_seconds
+        while cls._processed_thread_quote_cache:
+            oldest_key, oldest_ts = next(iter(cls._processed_thread_quote_cache.items()))
+            if oldest_ts >= expire_before:
+                break
+            cls._processed_thread_quote_cache.pop(oldest_key, None)
+
+        while len(cls._processed_thread_quote_cache) > cls._processed_thread_quote_cache_max_size:
+            oldest_key = next(iter(cls._processed_thread_quote_cache))
+            cls._processed_thread_quote_cache.pop(oldest_key, None)
+
+    @classmethod
+    def _mark_thread_quote_processed(cls, thread_id: str) -> None:
+        now = time.time()
+        cls._prune_processed_thread_quote_cache(now)
+        cls._processed_thread_quote_cache[thread_id] = now
+
+    @classmethod
+    def _extract_quote_message_id(cls, message: EventMessage) -> typing.Optional[str]:
+        """
+        Extract the message ID to quote from the given message.
+
+        Rules:
+        - First thread reply in a topic: return parent_id and mark topic as processed
+        - Follow-up thread replies in the same topic: return None
+        - Non-thread message: return parent_id if valid (non-empty, different from message_id)
+
+        Thread reply state is kept in a bounded TTL cache to avoid unbounded memory growth.
+        """
+        parent_id = getattr(message, 'parent_id', None)
+        if not parent_id:
+            return None
+
+        message_id = getattr(message, 'message_id', None)
+        if parent_id == message_id:
+            return None
+
+        thread_id = getattr(message, 'thread_id', None)
+        if thread_id:
+            cls._prune_processed_thread_quote_cache()
+            if thread_id in cls._processed_thread_quote_cache:
+                return None
+            cls._mark_thread_quote_processed(thread_id)
+
+        return parent_id
+
+    @staticmethod
+    def _build_event_message_from_message_item(message_item: Message) -> typing.Optional[EventMessage]:
+        """
+        Build EventMessage from SDK typed Message item.
+
+        Returns None if body or content is missing.
+        """
+        body = getattr(message_item, 'body', None)
+        if not body:
+            return None
+
+        content = getattr(body, 'content', None)
+        if not content:
+            return None
+
+        event_data = {
+            'message_id': message_item.message_id,
+            'message_type': message_item.msg_type,
+            'content': content,
+            'create_time': message_item.create_time,
+            'mentions': getattr(message_item, 'mentions', []) or [],
+        }
+
+        # Preserve thread-related fields
+        if hasattr(message_item, 'parent_id') and message_item.parent_id:
+            event_data['parent_id'] = message_item.parent_id
+        if hasattr(message_item, 'root_id') and message_item.root_id:
+            event_data['root_id'] = message_item.root_id
+        if hasattr(message_item, 'thread_id') and message_item.thread_id:
+            event_data['thread_id'] = message_item.thread_id
+        if hasattr(message_item, 'chat_id') and message_item.chat_id:
+            event_data['chat_id'] = message_item.chat_id
+
+        return EventMessage(event_data)
+
+    @staticmethod
+    async def _fetch_quoted_message(
+        quote_message_id: str,
+        api_client: lark_oapi.Client,
+    ) -> typing.Optional[platform_message.MessageChain]:
+        """
+        Fetch the quoted message and convert to MessageChain.
+
+        Returns None if:
+        - API call fails
+        - Response items is empty
+        - Message item normalization fails
+        """
+        request = GetMessageRequest.builder().message_id(quote_message_id).build()
+        response = await api_client.im.v1.message.aget(request)
+
+        if not response.success():
+            return None
+
+        items = getattr(response.data, 'items', None)
+        if not items:
+            return None
+
+        message_item = items[0]
+        event_message = LarkEventConverter._build_event_message_from_message_item(message_item)
+        if event_message is None:
+            return None
+
+        quote_chain = await LarkMessageConverter.target2yiri(event_message, api_client)
+        return quote_chain
+
     @staticmethod
     async def yiri2target(
         event: platform_events.MessageEvent,
@@ -587,6 +708,23 @@ class LarkEventConverter(abstract_platform_adapter.AbstractEventConverter):
     ) -> platform_events.Event:
         message_chain = await LarkMessageConverter.target2yiri(event.event.message, api_client)
 
+        # Check for quote/reply message
+        quote_message_id = LarkEventConverter._extract_quote_message_id(event.event.message)
+        if quote_message_id:
+            quote_chain = await LarkEventConverter._fetch_quoted_message(quote_message_id, api_client)
+            if quote_chain:
+                # Filter out Source component from quoted chain, keep only content
+                quote_origin = platform_message.MessageChain(
+                    [comp for comp in quote_chain if not isinstance(comp, platform_message.Source)]
+                )
+                if quote_origin:
+                    message_chain.append(
+                        platform_message.Quote(
+                            message_id=quote_message_id,
+                            origin=quote_origin,
+                        )
+                    )
+
         if event.event.message.chat_type == 'p2p':
             return platform_events.FriendMessage(
                 sender=platform_entities.Friend(
@@ -770,6 +908,32 @@ class LarkAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter):
             self.request_tenant_access_token(tenant_key)
         return self.tenant_access_tokens.get(tenant_key)['token'] if self.tenant_access_tokens.get(tenant_key) else None
 
+    def get_launcher_id(self, event: platform_events.MessageEvent) -> str | None:
+        """
+        Get topic-scoped launcher_id for thread-aware session isolation.
+
+        For group thread messages, returns "{group_id}_{thread_id}"
+        to ensure conversation context stays stable per topic.
+
+        Returns None for non-thread messages or P2P messages.
+        """
+        source_event = getattr(event.source_platform_object, 'event', None)
+        if not source_event:
+            return None
+
+        message = getattr(source_event, 'message', None)
+        if not message:
+            return None
+
+        thread_id = getattr(message, 'thread_id', None)
+        if not thread_id:
+            return None
+
+        if isinstance(event, platform_events.GroupMessage):
+            return f'{event.group.id}_{thread_id}'
+
+        return None
+
     def build_api_client(self, config):
         app_id = config['app_id']
         app_secret = config['app_secret']

src/langbot/pkg/platform/sources/legacy/gewechat.py

@@ -9,7 +9,7 @@ import copy
 import threading
 
 import quart
-import aiohttp
+from langbot.pkg.utils import httpclient
 
 import langbot_plugin.api.definition.abstract.platform.adapter as abstract_platform_adapter
 from ....core import app
@@ -639,14 +639,14 @@ class GeWeChatAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter):
 
     async def run_async(self):
         if not self.config['token']:
-            async with aiohttp.ClientSession() as session:
-                async with session.post(
-                    f'{self.config["gewechat_url"]}/v2/api/tools/getTokenId',
-                    json={'app_id': self.config['app_id']},
-                ) as response:
-                    if response.status != 200:
-                        raise Exception(f'获取gewechat token失败: {await response.text()}')
-                    self.config['token'] = (await response.json())['data']
+            session = httpclient.get_session()
+            async with session.post(
+                f'{self.config["gewechat_url"]}/v2/api/tools/getTokenId',
+                json={'app_id': self.config['app_id']},
+            ) as response:
+                if response.status != 200:
+                    raise Exception(f'获取gewechat token失败: {await response.text()}')
+                self.config['token'] = (await response.json())['data']
 
         self.bot = gewechat_client.GewechatClient(f'{self.config["gewechat_url"]}/v2/api', self.config['token'])
 

src/langbot/pkg/platform/sources/telegram.py

@@ -1,4 +1,5 @@
 from __future__ import annotations
+import time
 
 
 import telegram
@@ -9,9 +10,9 @@ import telegramify_markdown
 import typing
 import traceback
 import base64
-import aiohttp
 import pydantic
 
+from langbot.pkg.utils import httpclient
 import langbot_plugin.api.definition.abstract.platform.adapter as abstract_platform_adapter
 import langbot_plugin.api.entities.builtin.platform.message as platform_message
 import langbot_plugin.api.entities.builtin.platform.events as platform_events
@@ -33,9 +34,9 @@ class TelegramMessageConverter(abstract_platform_adapter.AbstractMessageConverte
                 if component.base64:
                     photo_bytes = base64.b64decode(component.base64)
                 elif component.url:
-                    async with aiohttp.ClientSession() as session:
-                        async with session.get(component.url) as response:
-                            photo_bytes = await response.read()
+                    session = httpclient.get_session()
+                    async with session.get(component.url) as response:
+                        photo_bytes = await response.read()
                 elif component.path:
                     with open(component.path, 'rb') as f:
                         photo_bytes = f.read()
@@ -74,10 +75,9 @@ class TelegramMessageConverter(abstract_platform_adapter.AbstractMessageConverte
             file_bytes = None
             file_format = ''
 
-            async with aiohttp.ClientSession(trust_env=True) as session:
-                async with session.get(file.file_path) as response:
-                    file_bytes = await response.read()
-                    file_format = 'image/jpeg'
+            async with httpclient.get_session(trust_env=True).get(file.file_path) as response:
+                file_bytes = await response.read()
+                file_format = 'image/jpeg'
 
             message_components.append(
                 platform_message.Image(
@@ -94,9 +94,8 @@ class TelegramMessageConverter(abstract_platform_adapter.AbstractMessageConverte
             file_bytes = None
             file_format = message.voice.mime_type or 'audio/ogg'
 
-            async with aiohttp.ClientSession(trust_env=True) as session:
-                async with session.get(file.file_path) as response:
-                    file_bytes = await response.read()
+            async with httpclient.get_session(trust_env=True).get(file.file_path) as response:
+                file_bytes = await response.read()
 
             message_components.append(
                 platform_message.Voice(
@@ -194,7 +193,31 @@ class TelegramAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter):
         )
 
     async def send_message(self, target_type: str, target_id: str, message: platform_message.MessageChain):
-        pass
+        components = await TelegramMessageConverter.yiri2target(message, self.bot)
+
+        chat_id_str, _, thread_id_str = str(target_id).partition('#')
+        chat_id: int | str = int(chat_id_str) if chat_id_str.lstrip('-').isdigit() else chat_id_str
+        message_thread_id = int(thread_id_str) if thread_id_str and thread_id_str.isdigit() else None
+
+        for component in components:
+            component_type = component.get('type')
+            args = {'chat_id': chat_id}
+            if message_thread_id is not None:
+                args['message_thread_id'] = message_thread_id
+
+            if component_type == 'text':
+                text = component.get('text', '')
+                if self.config['markdown_card'] is True:
+                    text = telegramify_markdown.markdownify(content=text)
+                    args['parse_mode'] = 'MarkdownV2'
+                args['text'] = text
+                await self.bot.send_message(**args)
+            elif component_type == 'photo':
+                photo = component.get('photo')
+                if photo is None:
+                    continue
+                args['photo'] = telegram.InputFile(photo)
+                await self.bot.send_photo(**args)
 
     async def reply_message(
         self,
@@ -228,6 +251,39 @@ class TelegramAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter):
 
         await self.bot.send_message(**args)
 
+    def _process_markdown(self, text: str) -> str:
+        if self.config.get('markdown_card', False):
+            return telegramify_markdown.markdownify(content=text)
+        return text
+
+    def _build_message_args(self, chat_id: int, text: str, message_thread_id: int = None, **extra_args) -> dict:
+        args = {'chat_id': chat_id, 'text': self._process_markdown(text), **extra_args}
+        if message_thread_id:
+            args['message_thread_id'] = message_thread_id
+        if self.config.get('markdown_card', False):
+            args['parse_mode'] = 'MarkdownV2'
+        return args
+
+    async def create_message_card(self, message_id, event):
+        assert isinstance(event.source_platform_object, Update)
+        update = event.source_platform_object
+        chat_id = update.effective_chat.id
+        chat_type = update.effective_chat.type
+        message_thread_id = update.message.message_thread_id
+
+        if chat_type == 'private':
+            draft_id = int(time.time() * 1000)
+            self.msg_stream_id[message_id] = ('private', draft_id)
+
+            args = self._build_message_args(chat_id, 'Thinking...', message_thread_id, draft_id=draft_id)
+            await self.bot.send_message_draft(**args)
+        else:
+            args = self._build_message_args(chat_id, 'Thinking...', message_thread_id)
+            send_msg = await self.bot.send_message(**args)
+            self.msg_stream_id[message_id] = ('group', send_msg.message_id)
+
+        return True
+
     async def reply_message_chunk(
         self,
         message_source: platform_events.MessageEvent,
@@ -236,59 +292,47 @@ class TelegramAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter):
         quote_origin: bool = False,
         is_final: bool = False,
     ):
+        message_id = bot_message.resp_message_id
         msg_seq = bot_message.msg_sequence
-        if (msg_seq - 1) % 8 == 0 or is_final:
-            assert isinstance(message_source.source_platform_object, Update)
-            components = await TelegramMessageConverter.yiri2target(message, self.bot)
-            args = {}
-            message_id = message_source.source_platform_object.message.id
+        assert isinstance(message_source.source_platform_object, Update)
+        update = message_source.source_platform_object
+        chat_id = update.effective_chat.id
+        message_thread_id = update.message.message_thread_id
 
-            component = components[0]
-            if message_id not in self.msg_stream_id:  # 当消息回复第一次时，发送新消息
-                # time.sleep(0.6)
-                if component['type'] == 'text':
-                    if self.config['markdown_card'] is True:
-                        content = telegramify_markdown.markdownify(
-                            content=component['text'],
-                        )
-                    else:
-                        content = component['text']
-                    args = {
-                        'chat_id': message_source.source_platform_object.effective_chat.id,
-                        'text': content,
-                    }
-                    if message_source.source_platform_object.message.message_thread_id:
-                        args['message_thread_id'] = message_source.source_platform_object.message.message_thread_id
+        if message_id not in self.msg_stream_id:
+            return
 
-                    if quote_origin:
-                        args['reply_to_message_id'] = message_source.source_platform_object.message.id
+        chat_mode, draft_id = self.msg_stream_id[message_id]
+        components = await TelegramMessageConverter.yiri2target(message, self.bot)
 
-                    if self.config['markdown_card'] is True:
-                        args['parse_mode'] = 'MarkdownV2'
-
-                send_msg = await self.bot.send_message(**args)
-                send_msg_id = send_msg.message_id
-                self.msg_stream_id[message_id] = send_msg_id
-            else:  # 存在消息的时候直接编辑消息1
-                if component['type'] == 'text':
-                    if self.config['markdown_card'] is True:
-                        content = telegramify_markdown.markdownify(
-                            content=component['text'],
-                        )
-                    else:
-                        content = component['text']
-                    args = {
-                        'message_id': self.msg_stream_id[message_id],
-                        'chat_id': message_source.source_platform_object.effective_chat.id,
-                        'text': content,
-                    }
-                    if self.config['markdown_card'] is True:
-                        args['parse_mode'] = 'MarkdownV2'
-
-                await self.bot.edit_message_text(**args)
+        if not components or components[0]['type'] != 'text':
             if is_final and bot_message.tool_calls is None:
-                # self.seq = 1  # 消息回复结束之后重置seq
-                self.msg_stream_id.pop(message_id)  # 消息回复结束之后删除流式消息id
+                self.msg_stream_id.pop(message_id)
+            return
+
+        content = components[0]['text']
+
+        if chat_mode == 'private':
+            args = self._build_message_args(chat_id, content, message_thread_id, draft_id=draft_id)
+            await self.bot.send_message_draft(**args)
+            if is_final and bot_message.tool_calls is None:
+                del args['draft_id']
+                await self.bot.send_message(**args)
+                self.msg_stream_id.pop(message_id)
+        else:
+            stream_id = draft_id
+            if (msg_seq - 1) % 8 == 0 or is_final:
+                args = {
+                    'message_id': stream_id,
+                    'chat_id': chat_id,
+                    'text': self._process_markdown(content),
+                }
+                if self.config.get('markdown_card', False):
+                    args['parse_mode'] = 'MarkdownV2'
+                await self.bot.edit_message_text(**args)
+
+            if is_final and bot_message.tool_calls is None:
+                self.msg_stream_id.pop(message_id)
 
     def get_launcher_id(self, event: platform_events.MessageEvent) -> str | None:
         if not isinstance(event.source_platform_object, Update):

src/langbot/pkg/platform/sources/websocket_adapter.py

@@ -37,16 +37,24 @@ class WebSocketSession:
     id: str
     message_lists: dict[str, list[WebSocketMessage]] = {}
     """消息列表 {pipeline_uuid: [messages]}"""
+    stream_message_indexes: dict[str, dict[str, int]] = {}
+    """流式消息索引 {pipeline_uuid: {resp_message_id: message_index}}"""
 
     def __init__(self, id: str):
         self.id = id
         self.message_lists = {}
+        self.stream_message_indexes = {}
 
     def get_message_list(self, pipeline_uuid: str) -> list[WebSocketMessage]:
         if pipeline_uuid not in self.message_lists:
             self.message_lists[pipeline_uuid] = []
         return self.message_lists[pipeline_uuid]
 
+    def get_stream_message_indexes(self, pipeline_uuid: str) -> dict[str, int]:
+        if pipeline_uuid not in self.stream_message_indexes:
+            self.stream_message_indexes[pipeline_uuid] = {}
+        return self.stream_message_indexes[pipeline_uuid]
+
 
 class WebSocketAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter):
     """WebSocket适配器 - 支持双向实时通信"""
@@ -89,20 +97,46 @@ class WebSocketAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter)
         target_id: str,
         message: platform_message.MessageChain,
     ) -> dict:
-        """发送消息 - 这里用于主动推送消息到前端"""
-        message_data = {
-            'type': 'bot_message',
-            'target_type': target_type,
-            'target_id': target_id,
-            'content': str(message),
-            'message_chain': [component.__dict__ for component in message],
-            'timestamp': datetime.now().isoformat(),
-        }
+        """发送消息 - 这里用于主动推送消息到前端
 
-        # 推送到所有相关连接
-        await self.outbound_message_queue.put(message_data)
+        对于 WebSocket 适配器，我们需要将消息广播到正确的 pipeline 连接。
+        target_id 可能是 launcher_id（如 websocket_xxx）或 pipeline_uuid。
+        我们需要尝试两种方式来确保消息能够送达。
+        """
+        # 获取当前的 pipeline_uuid
+        pipeline_uuid = self.ap.platform_mgr.websocket_proxy_bot.bot_entity.use_pipeline_uuid
+        session_type = 'group' if target_type == 'group' else 'person'
 
-        return message_data
+        # 选择会话
+        session = self.websocket_group_session if session_type == 'group' else self.websocket_person_session
+
+        # 生成唯一消息ID
+        msg_id = len(session.get_message_list(pipeline_uuid)) + 1
+
+        message_data = WebSocketMessage(
+            id=msg_id,
+            role='assistant',
+            content=str(message),
+            message_chain=[component.__dict__ for component in message],
+            timestamp=datetime.now().isoformat(),
+            is_final=True,
+        )
+
+        # 保存到历史记录
+        session.get_message_list(pipeline_uuid).append(message_data)
+
+        # 直接广播到当前pipeline的连接
+        await ws_connection_manager.broadcast_to_pipeline(
+            pipeline_uuid,
+            {
+                'type': 'response',
+                'session_type': session_type,
+                'data': message_data.model_dump(),
+            },
+            session_type=session_type,
+        )
+
+        return message_data.model_dump()
 
     async def reply_message(
         self,
@@ -169,10 +203,16 @@ class WebSocketAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter)
         pipeline_uuid = self.ap.platform_mgr.websocket_proxy_bot.bot_entity.use_pipeline_uuid
         session_type = 'group' if isinstance(message_source, platform_events.GroupMessage) else 'person'
         message_list = session.get_message_list(pipeline_uuid)
+        stream_message_indexes = session.get_stream_message_indexes(pipeline_uuid)
 
-        # 检查是否是新的流式消息（通过bot_message对象判断）
-        # 如果列表为空，或者最后一条消息已经is_final=True，则创建新消息
-        if not message_list or message_list[-1].is_final:
+        # Streaming messages in LangBot have a stable resp_message_id during the same assistant reply.
+        # Use it as the primary key to avoid overwriting an old card from a previous reply.
+        resp_message_id = str(getattr(bot_message, 'resp_message_id', '') or '')
+        existing_index = stream_message_indexes.get(resp_message_id) if resp_message_id else None
+
+        message_is_final = is_final and bot_message.tool_calls is None
+
+        if existing_index is None or existing_index >= len(message_list):
             # 创建新消息
             msg_id = len(message_list) + 1
             message_data = WebSocketMessage(
@@ -181,27 +221,31 @@ class WebSocketAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter)
                 content=str(message),
                 message_chain=[component.__dict__ for component in message],
                 timestamp=datetime.now().isoformat(),
-                is_final=is_final and bot_message.tool_calls is None,
+                is_final=message_is_final,
             )
 
-            # 只有在is_final时才保存到历史记录
-            if is_final and bot_message.tool_calls is None:
-                message_list.append(message_data)
+            # 立即添加到历史记录（即使is_final=False），以便后续块可以更新它
+            message_list.append(message_data)
+            if resp_message_id:
+                stream_message_indexes[resp_message_id] = len(message_list) - 1
         else:
-            # 更新最后一条消息
-            msg_id = message_list[-1].id
+            # 更新同一条流式消息
+            old_message = message_list[existing_index]
+            msg_id = old_message.id
             message_data = WebSocketMessage(
                 id=msg_id,
                 role='assistant',
                 content=str(message),
                 message_chain=[component.__dict__ for component in message],
-                timestamp=message_list[-1].timestamp,  # 保持原始时间戳
-                is_final=is_final and bot_message.tool_calls is None,
+                timestamp=old_message.timestamp,  # 保持原始时间戳
+                is_final=message_is_final,
             )
 
-            # 如果是final，更新历史记录中的最后一条
-            if is_final and bot_message.tool_calls is None:
-                message_list[-1] = message_data
+            # 更新历史记录中的对应消息
+            message_list[existing_index] = message_data
+
+        if message_is_final and resp_message_id:
+            stream_message_indexes.pop(resp_message_id, None)
 
         # 直接广播到所有该pipeline的连接，包含session_type信息
         await ws_connection_manager.broadcast_to_pipeline(
@@ -410,6 +454,10 @@ class WebSocketAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter)
         if session_type == 'person':
             if pipeline_uuid in self.websocket_person_session.message_lists:
                 self.websocket_person_session.message_lists[pipeline_uuid] = []
+            if pipeline_uuid in self.websocket_person_session.stream_message_indexes:
+                self.websocket_person_session.stream_message_indexes[pipeline_uuid] = {}
         else:
             if pipeline_uuid in self.websocket_group_session.message_lists:
                 self.websocket_group_session.message_lists[pipeline_uuid] = []
+            if pipeline_uuid in self.websocket_group_session.stream_message_indexes:
+                self.websocket_group_session.stream_message_indexes[pipeline_uuid] = {}

src/langbot/pkg/platform/sources/wecombot.py

@@ -11,6 +11,7 @@ import langbot_plugin.api.entities.builtin.platform.entities as platform_entitie
 from ..logger import EventLogger
 from langbot.libs.wecom_ai_bot_api.wecombotevent import WecomBotEvent
 from langbot.libs.wecom_ai_bot_api.api import WecomBotClient
+from langbot.libs.wecom_ai_bot_api.ws_client import WecomBotWsClient
 
 
 class WecomBotMessageConverter(abstract_platform_adapter.AbstractMessageConverter):
@@ -176,27 +177,42 @@ class WecomBotEventConverter(abstract_platform_adapter.AbstractEventConverter):
 
 
 class WecomBotAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter):
-    bot: WecomBotClient
+    bot: typing.Union[WecomBotClient, WecomBotWsClient]
     bot_account_id: str
     message_converter: WecomBotMessageConverter = WecomBotMessageConverter()
     event_converter: WecomBotEventConverter = WecomBotEventConverter()
     config: dict
     bot_uuid: str = None
+    _ws_mode: bool = False
 
     def __init__(self, config: dict, logger: EventLogger):
-        required_keys = ['Token', 'EncodingAESKey', 'Corpid', 'BotId']
-        missing_keys = [key for key in required_keys if key not in config]
-        if missing_keys:
-            raise Exception(f'WecomBot 缺少配置项: {missing_keys}')
+        enable_webhook = config.get('enable-webhook', False)
 
-        bot = WecomBotClient(
-            Token=config['Token'],
-            EnCodingAESKey=config['EncodingAESKey'],
-            Corpid=config['Corpid'],
-            logger=logger,
-            unified_mode=True,
-        )
-        bot_account_id = config['BotId']
+        if not enable_webhook:
+            bot = WecomBotWsClient(
+                bot_id=config['BotId'],
+                secret=config['Secret'],
+                logger=logger,
+                encoding_aes_key=config.get('EncodingAESKey', ''),
+            )
+            ws_mode = True
+        else:
+            # Webhook callback mode
+            required_keys = ['Token', 'EncodingAESKey', 'Corpid']
+            missing_keys = [key for key in required_keys if key not in config or not config[key]]
+            if missing_keys:
+                raise Exception(f'WecomBot webhook mode missing config: {missing_keys}')
+
+            bot = WecomBotClient(
+                Token=config['Token'],
+                EnCodingAESKey=config['EncodingAESKey'],
+                Corpid=config['Corpid'],
+                logger=logger,
+                unified_mode=True,
+            )
+            ws_mode = False
+
+        bot_account_id = config.get('BotId', '')
 
         super().__init__(
             config=config,
@@ -204,6 +220,7 @@ class WecomBotAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter):
             bot=bot,
             bot_account_id=bot_account_id,
         )
+        self._ws_mode = ws_mode
 
     async def reply_message(
         self,
@@ -212,7 +229,15 @@ class WecomBotAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter):
         quote_origin: bool = False,
     ):
         content = await self.message_converter.yiri2target(message)
-        await self.bot.set_message(message_source.source_platform_object.message_id, content)
+        if self._ws_mode:
+            event = message_source.source_platform_object
+            req_id = event.get('req_id', '')
+            if req_id:
+                await self.bot.reply_text(req_id, content)
+            else:
+                await self.bot.set_message(event.message_id, content)
+        else:
+            await self.bot.set_message(message_source.source_platform_object.message_id, content)
 
     async def reply_message_chunk(
         self,
@@ -222,31 +247,22 @@ class WecomBotAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter):
         quote_origin: bool = False,
         is_final: bool = False,
     ):
-        """将流水线增量输出写入企业微信 stream 会话。
-
-        Args:
-            message_source: 流水线提供的原始消息事件。
-            bot_message: 当前片段对应的模型元信息（未使用）。
-            message: 需要回复的消息链。
-            quote_origin: 是否引用原消息（企业微信暂不支持）。
-            is_final: 标记当前片段是否为最终回复。
-
-        Returns:
-            dict: 包含 `stream` 键，标识写入是否成功。
-
-        Example:
-            在流水线 `reply_message_chunk` 调用中自动触发，无需手动调用。
-        """
-        # 转换为纯文本（智能机器人当前协议仅支持文本流）
         content = await self.message_converter.yiri2target(message)
         msg_id = message_source.source_platform_object.message_id
 
-        # 将片段推送到 WecomBotClient 中的队列，返回值用于判断是否走降级逻辑
-        success = await self.bot.push_stream_chunk(msg_id, content, is_final=is_final)
-        if not success and is_final:
-            # 未命中流式队列时使用旧有 set_message 兜底
-            await self.bot.set_message(msg_id, content)
-        return {'stream': success}
+        if self._ws_mode:
+            success = await self.bot.push_stream_chunk(msg_id, content, is_final=is_final)
+            if not success and is_final:
+                event = message_source.source_platform_object
+                req_id = event.get('req_id', '')
+                if req_id:
+                    await self.bot.reply_text(req_id, content)
+            return {'stream': success}
+        else:
+            success = await self.bot.push_stream_chunk(msg_id, content, is_final=is_final)
+            if not success and is_final:
+                await self.bot.set_message(msg_id, content)
+            return {'stream': success}
 
     async def is_stream_output_supported(self) -> bool:
         """智能机器人侧默认开启流式能力。
@@ -259,7 +275,11 @@ class WecomBotAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter):
         return True
 
     async def send_message(self, target_type, target_id, message):
-        pass
+        if self._ws_mode:
+            content = await self.message_converter.yiri2target(message)
+            await self.bot.send_message(target_id, content)
+        else:
+            pass
 
     def register_listener(
         self,
@@ -288,29 +308,25 @@ class WecomBotAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter):
         self.bot_uuid = bot_uuid
 
     async def handle_unified_webhook(self, bot_uuid: str, path: str, request):
-        """处理统一 webhook 请求。
-
-        Args:
-            bot_uuid: Bot 的 UUID
-            path: 子路径（如果有的话）
-            request: Quart Request 对象
-
-        Returns:
-            响应数据
-        """
+        if self._ws_mode:
+            return None
         return await self.bot.handle_unified_webhook(request)
 
     async def run_async(self):
-        # 统一 webhook 模式下，不启动独立的 Quart 应用
-        # 保持运行但不启动独立端口
+        if self._ws_mode:
+            await self.bot.connect()
+        else:
 
-        async def keep_alive():
-            while True:
-                await asyncio.sleep(1)
+            async def keep_alive():
+                while True:
+                    await asyncio.sleep(1)
 
-        await keep_alive()
+            await keep_alive()
 
     async def kill(self) -> bool:
+        if self._ws_mode:
+            await self.bot.disconnect()
+            return True
         return False
 
     async def unregister_listener(

src/langbot/pkg/platform/sources/wecombot.yaml

@@ -11,35 +11,64 @@ metadata:
   icon: wecombot.png
 spec:
   config:
+    - name: BotId
+      label:
+        en_US: BotId
+        zh_Hans: 机器人ID (BotId)
+      type: string
+      required: true
+      default: ""
+    - name: enable-webhook
+      label:
+        en_US: Enable Webhook Mode
+        zh_Hans: 启用Webhook模式
+      description:
+        en_US: If enabled, the bot will use webhook mode to receive messages. Otherwise, it will use WS long connection mode
+        zh_Hans: 如果启用，机器人将使用 Webhook 模式接收消息。否则，将使用 WS 长连接模式
+      type: boolean
+      required: true
+      default: false
+    - name: Secret
+      label:
+        en_US: Secret
+        zh_Hans: 机器人密钥 (Secret)
+      description:
+        en_US: Required for WebSocket long connection mode
+        zh_Hans: 使用 WS 长连接模式时必填
+      type: string
+      required: false
+      default: ""
     - name: Corpid
       label:
         en_US: Corpid
         zh_Hans: 企业ID
+      description:
+        en_US: Required for Webhook mode
+        zh_Hans: 使用 Webhook 模式时必填
       type: string
-      required: true
+      required: false
       default: ""
     - name: Token
       label:
         en_US: Token
         zh_Hans: 令牌 (Token)
+      description:
+        en_US: Required for Webhook mode
+        zh_Hans: 使用 Webhook 模式时必填
       type: string
-      required: true
+      required: false
       default: ""
     - name: EncodingAESKey
       label:
         en_US: EncodingAESKey
         zh_Hans: 消息加解密密钥 (EncodingAESKey)
-      type: string
-      required: true
-      default: ""
-    - name: BotId
-      label:
-        en_US: BotId
-        zh_Hans: 机器人ID
+      description:
+        en_US: Required for Webhook mode. Optional for WebSocket mode (used for file decryption)
+        zh_Hans: 使用 Webhook 模式时必填。WebSocket 模式下可选（用于文件解密）
       type: string
       required: false
       default: ""
 execution:
   python:
     path: ./wecombot.py
-    attr: WecomBotAdapter
+    attr: WecomBotAdapter

src/langbot/pkg/platform/sources/wecomcs.py

@@ -81,22 +81,33 @@ class WecomEventConverter(abstract_platform_adapter.AbstractEventConverter):
             return event.source_platform_object
 
     @staticmethod
-    async def target2yiri(event: WecomCSEvent):
+    async def target2yiri(event: WecomCSEvent, bot: WecomCSClient = None):
         """
         将 WecomEvent 转换为平台的 FriendMessage 对象。
 
         Args:
             event (WecomEvent): 企业微信客服事件。
+            bot (WecomCSClient): 企业微信客服客户端，用于获取用户信息。
 
         Returns:
             platform_events.FriendMessage: 转换后的 FriendMessage 对象。
         """
+        # Try to get customer nickname from WeChat API
+        nickname = str(event.user_id)
+        if bot and event.user_id:
+            try:
+                customer_info = await bot.get_customer_info(event.user_id)
+                if customer_info and customer_info.get('nickname'):
+                    nickname = customer_info.get('nickname')
+            except Exception:
+                pass  # Fall back to user_id as nickname
+
         # 转换消息链
         if event.type == 'text':
             yiri_chain = await WecomMessageConverter.target2yiri(event.message, event.message_id)
             friend = platform_entities.Friend(
                 id=f'u{event.user_id}',
-                nickname=str(event.user_id),
+                nickname=nickname,
                 remark='',
             )
 
@@ -106,7 +117,7 @@ class WecomEventConverter(abstract_platform_adapter.AbstractEventConverter):
         elif event.type == 'image':
             friend = platform_entities.Friend(
                 id=f'u{event.user_id}',
-                nickname=str(event.user_id),
+                nickname=nickname,
                 remark='',
             )
 
@@ -187,7 +198,7 @@ class WecomCSAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter):
         async def on_message(event: WecomCSEvent):
             self.bot_account_id = event.receiver_id
             try:
-                return await callback(await self.event_converter.target2yiri(event), self)
+                return await callback(await self.event_converter.target2yiri(event, self.bot), self)
             except Exception:
                 await self.logger.error(f'Error in wecomcs callback: {traceback.format_exc()}')
 

src/langbot/pkg/platform/webhook_pusher.py

@@ -3,6 +3,8 @@ from __future__ import annotations
 import asyncio
 import logging
 import aiohttp
+
+from langbot.pkg.utils import httpclient
 import uuid
 from typing import TYPE_CHECKING
 
@@ -119,23 +121,23 @@ class WebhookPusher:
             dict | None: The response JSON if successful, None otherwise
         """
         try:
-            async with aiohttp.ClientSession() as session:
-                async with session.post(
-                    url,
-                    json=payload,
-                    headers={'Content-Type': 'application/json'},
-                    timeout=aiohttp.ClientTimeout(total=15),
-                ) as response:
-                    if response.status >= 400:
-                        self.logger.warning(f'Webhook {url} returned status {response.status}')
+            session = httpclient.get_session()
+            async with session.post(
+                url,
+                json=payload,
+                headers={'Content-Type': 'application/json'},
+                timeout=aiohttp.ClientTimeout(total=15),
+            ) as response:
+                if response.status >= 400:
+                    self.logger.warning(f'Webhook {url} returned status {response.status}')
+                    return None
+                else:
+                    self.logger.debug(f'Successfully pushed to webhook {url}')
+                    try:
+                        return await response.json()
+                    except Exception as json_error:
+                        self.logger.debug(f'Failed to parse JSON response from webhook {url}: {json_error}')
                         return None
-                    else:
-                        self.logger.debug(f'Successfully pushed to webhook {url}')
-                        try:
-                            return await response.json()
-                        except Exception as json_error:
-                            self.logger.debug(f'Failed to parse JSON response from webhook {url}: {json_error}')
-                            return None
         except asyncio.TimeoutError:
             self.logger.warning(f'Timeout pushing to webhook {url}')
             return None

src/langbot/pkg/plugin/connector.py

@@ -7,7 +7,6 @@ import typing
 import os
 import sys
 import httpx
-import traceback
 import sqlalchemy
 from async_lru import alru_cache
 from langbot_plugin.api.entities.builtin.pipeline.query import provider_session
@@ -102,12 +101,6 @@ class PluginRuntimeConnector:
             self.handler_task = asyncio.create_task(self.handler.run())
             _ = await self.handler.ping()
             self.ap.logger.info('Connected to plugin runtime.')
-            # Sync polymorphic component instances after connection
-            try:
-                await self.sync_polymorphic_component_instances()
-            except Exception as e:
-                traceback.print_exc()
-                self.ap.logger.error(f'Failed to sync polymorphic component instances: {e}')
             await self.handler_task
 
         task: asyncio.Task | None = None
@@ -463,30 +456,18 @@ class PluginRuntimeConnector:
 
             yield cmd_ret
 
-    # KnowledgeRetriever methods
-    async def list_knowledge_retrievers(self, bound_plugins: list[str] | None = None) -> list[dict[str, Any]]:
-        """List all available KnowledgeRetriever components."""
-        if not self.is_enable_plugin:
-            return []
-
-        retrievers_data = await self.handler.list_knowledge_retrievers(include_plugins=bound_plugins)
-        return retrievers_data
-
     async def retrieve_knowledge(
         self,
         plugin_author: str,
         plugin_name: str,
         retriever_name: str,
-        instance_id: str,
         retrieval_context: dict[str, Any],
-    ) -> list[dict[str, Any]]:
-        """Retrieve knowledge using a KnowledgeRetriever instance."""
+    ) -> dict[str, Any]:
+        """Retrieve knowledge using a KnowledgeEngine instance."""
         if not self.is_enable_plugin:
-            return []
+            return {'results': []}
 
-        return await self.handler.retrieve_knowledge(
-            plugin_author, plugin_name, retriever_name, instance_id, retrieval_context
-        )
+        return await self.handler.retrieve_knowledge(plugin_author, plugin_name, retriever_name, retrieval_context)
 
     def dispose(self):
         # No need to consider the shutdown on Windows
@@ -500,41 +481,84 @@ class PluginRuntimeConnector:
             self.heartbeat_task.cancel()
             self.heartbeat_task = None
 
-    async def sync_polymorphic_component_instances(self) -> dict[str, Any]:
-        """Sync polymorphic component instances with runtime.
+    @staticmethod
+    def _parse_plugin_id(plugin_id: str) -> tuple[str, str]:
+        """Parse a plugin ID string into (author, name).
 
-        This collects all external knowledge bases from database and sends to runtime
-        to ensure instance integrity across restarts.
+        Args:
+            plugin_id: Plugin ID in 'author/name' format.
+
+        Returns:
+            Tuple of (plugin_author, plugin_name).
+
+        Raises:
+            ValueError: If plugin_id is not in the expected 'author/name' format.
+        """
+        if '/' not in plugin_id:
+            raise ValueError(
+                f"Invalid plugin_id format: '{plugin_id}'. Expected 'author/name' format (e.g. 'langbot/rag-engine')."
+            )
+        return plugin_id.split('/', 1)
+
+    async def call_rag_ingest(self, plugin_id: str, context_data: dict[str, Any]) -> dict[str, Any]:
+        """Call plugin to ingest document.
+
+        Args:
+            plugin_id: Target plugin ID (author/name).
+            context_data: IngestionContext data.
+        """
+        plugin_author, plugin_name = self._parse_plugin_id(plugin_id)
+        return await self.handler.rag_ingest_document(plugin_author, plugin_name, context_data)
+
+    async def call_rag_delete_document(self, plugin_id: str, document_id: str, kb_id: str) -> bool:
+        plugin_author, plugin_name = self._parse_plugin_id(plugin_id)
+        return await self.handler.rag_delete_document(plugin_author, plugin_name, document_id, kb_id)
+
+    async def get_rag_creation_schema(self, plugin_id: str) -> dict[str, Any]:
+        plugin_author, plugin_name = self._parse_plugin_id(plugin_id)
+        return await self.handler.get_rag_creation_schema(plugin_author, plugin_name)
+
+    async def get_rag_retrieval_schema(self, plugin_id: str) -> dict[str, Any]:
+        plugin_author, plugin_name = self._parse_plugin_id(plugin_id)
+        return await self.handler.get_rag_retrieval_schema(plugin_author, plugin_name)
+
+    async def rag_on_kb_create(self, plugin_id: str, kb_id: str, config: dict[str, Any]) -> dict[str, Any]:
+        """Notify plugin about KB creation."""
+        plugin_author, plugin_name = self._parse_plugin_id(plugin_id)
+        return await self.handler.rag_on_kb_create(plugin_author, plugin_name, kb_id, config)
+
+    async def rag_on_kb_delete(self, plugin_id: str, kb_id: str) -> dict[str, Any]:
+        """Notify plugin about KB deletion."""
+        plugin_author, plugin_name = self._parse_plugin_id(plugin_id)
+        return await self.handler.rag_on_kb_delete(plugin_author, plugin_name, kb_id)
+
+    async def call_rag_retrieve(self, plugin_id: str, retrieval_context: dict[str, Any]) -> dict[str, Any]:
+        """Call plugin to retrieve knowledge.
+
+        Args:
+            plugin_id: Target plugin ID (author/name).
+            retrieval_context: RetrievalContext data.
+        """
+        plugin_author, plugin_name = self._parse_plugin_id(plugin_id)
+        return await self.handler.retrieve_knowledge(plugin_author, plugin_name, '', retrieval_context)
+
+    async def list_knowledge_engines(self) -> list[dict[str, Any]]:
+        """List all available Knowledge Engines from plugins.
+
+        Returns a list of Knowledge Engines with their capabilities and configuration schemas.
         """
         if not self.is_enable_plugin:
-            return {}
+            return []
 
-        # ===== external knowledge bases =====
+        return await self.handler.list_knowledge_engines()
 
-        external_kbs = await self.ap.external_kb_service.get_external_knowledge_bases()
+    async def list_parsers(self) -> list[dict[str, Any]]:
+        """List all available parsers from plugins."""
+        if not self.is_enable_plugin:
+            return []
+        return await self.handler.list_parsers()
 
-        # Build required_instances list
-        required_instances = []
-        for kb in external_kbs:
-            required_instances.append(
-                {
-                    'instance_id': kb['uuid'],
-                    'plugin_author': kb['plugin_author'],
-                    'plugin_name': kb['plugin_name'],
-                    'component_kind': 'KnowledgeRetriever',
-                    'component_name': kb['retriever_name'],
-                    'config': kb['retriever_config'],
-                }
-            )
-
-        self.ap.logger.info(f'Syncing {len(required_instances)} polymorphic component instances to runtime')
-
-        # Send to runtime
-        sync_result = await self.handler.sync_polymorphic_component_instances(required_instances)
-
-        self.ap.logger.info(
-            f'Sync complete: {len(sync_result.get("success_instances", []))} succeeded, '
-            f'{len(sync_result.get("failed_instances", []))} failed'
-        )
-
-        return sync_result
+    async def call_parser(self, plugin_id: str, context_data: dict[str, Any], file_bytes: bytes) -> dict[str, Any]:
+        """Call plugin to parse a document."""
+        plugin_author, plugin_name = self._parse_plugin_id(plugin_id)
+        return await self.handler.parse_document(plugin_author, plugin_name, context_data, file_bytes)

src/langbot/pkg/plugin/handler.py

@@ -26,6 +26,20 @@ from ..core import app
 from ..utils import constants
 
 
+def _make_rag_error_response(error: Exception, error_type: str, **extra_context) -> handler.ActionResponse:
+    """Create a clean error response for RAG operations.
+
+    Args:
+        error: The caught exception.
+        error_type: A category string like 'EmbeddingError', 'VectorStoreError'.
+        **extra_context: Additional context fields for the error message.
+    """
+    context_parts = [f'{k}={v}' for k, v in extra_context.items()]
+    context_str = f' [{", ".join(context_parts)}]' if context_parts else ''
+    message = f'[{error_type}/{type(error).__name__}]{context_str} {str(error)}'
+    return handler.ActionResponse.error(message=message)
+
+
 class RuntimeConnectionHandler(handler.Handler):
     """Runtime connection handler"""
 
@@ -323,7 +337,14 @@ class RuntimeConnectionHandler(handler.Handler):
                 )
 
             messages_obj = [provider_message.Message.model_validate(message) for message in messages]
-            funcs_obj = [resource_tool.LLMTool.model_validate(func) for func in funcs]
+
+            # The func field is excluded during model_dump() in plugin side (marked as exclude=True),
+            # but it's a required field for LLMTool validation. We need to provide a placeholder
+            # function when reconstructing the LLMTool objects from serialized data.
+            async def _placeholder_func(**kwargs):
+                pass
+
+            funcs_obj = [resource_tool.LLMTool.model_validate({**func, 'func': _placeholder_func}) for func in funcs]
 
             result = await llm_model.provider.invoke_llm(
                 query=None,
@@ -439,7 +460,7 @@ class RuntimeConnectionHandler(handler.Handler):
                 },
             )
 
-        @self.action(RuntimeToLangBotAction.GET_CONFIG_FILE)
+        @self.action(PluginToRuntimeAction.GET_CONFIG_FILE)
         async def get_config_file(data: dict[str, Any]) -> handler.ActionResponse:
             """Get a config file by file key"""
             file_key = data['file_key']
@@ -458,6 +479,227 @@ class RuntimeConnectionHandler(handler.Handler):
                     message=f'Failed to load config file {file_key}: {e}',
                 )
 
+        # ================= RAG Capability Handlers =================
+
+        @self.action(PluginToRuntimeAction.INVOKE_EMBEDDING)
+        async def invoke_embedding(data: dict[str, Any]) -> handler.ActionResponse:
+            embedding_model_uuid = data['embedding_model_uuid']
+            texts = data['texts']
+
+            embedding_model = await self.ap.model_mgr.get_embedding_model_by_uuid(embedding_model_uuid)
+            if embedding_model is None:
+                return handler.ActionResponse.error(
+                    message=f'Embedding model with embedding_model_uuid {embedding_model_uuid} not found',
+                )
+
+            try:
+                vectors = await embedding_model.provider.invoke_embedding(embedding_model, texts)
+                return handler.ActionResponse.success(data={'vectors': vectors})
+            except Exception as e:
+                return _make_rag_error_response(e, 'EmbeddingError', embedding_model_uuid=embedding_model_uuid)
+
+        @self.action(PluginToRuntimeAction.VECTOR_UPSERT)
+        async def vector_upsert(data: dict[str, Any]) -> handler.ActionResponse:
+            collection_id = data['collection_id']
+            vectors = data['vectors']
+            ids = data['ids']
+            metadata = data.get('metadata')
+            documents = data.get('documents')
+            if len(vectors) != len(ids):
+                return handler.ActionResponse.error(message='vectors and ids must have same length')
+            if metadata and len(metadata) != len(vectors):
+                return handler.ActionResponse.error(message='metadata must match vectors length')
+            if documents and len(documents) != len(vectors):
+                return handler.ActionResponse.error(message='documents must match vectors length')
+            try:
+                await self.ap.rag_runtime_service.vector_upsert(
+                    collection_id,
+                    vectors,
+                    ids,
+                    metadata,
+                    documents,
+                )
+                return handler.ActionResponse.success(data={})
+            except Exception as e:
+                return _make_rag_error_response(e, 'VectorStoreError', collection_id=collection_id)
+
+        @self.action(PluginToRuntimeAction.VECTOR_SEARCH)
+        async def vector_search(data: dict[str, Any]) -> handler.ActionResponse:
+            collection_id = data['collection_id']
+            query_vector = data['query_vector']
+            top_k = data['top_k']
+            filters = data.get('filters')
+            search_type = data.get('search_type', 'vector')
+            query_text = data.get('query_text', '')
+            try:
+                results = await self.ap.rag_runtime_service.vector_search(
+                    collection_id,
+                    query_vector,
+                    top_k,
+                    filters,
+                    search_type,
+                    query_text,
+                )
+                return handler.ActionResponse.success(data={'results': results})
+            except Exception as e:
+                return _make_rag_error_response(e, 'VectorStoreError', collection_id=collection_id)
+
+        @self.action(PluginToRuntimeAction.VECTOR_DELETE)
+        async def vector_delete(data: dict[str, Any]) -> handler.ActionResponse:
+            collection_id = data['collection_id']
+            file_ids = data.get('file_ids')
+            filters = data.get('filters')
+            try:
+                count = await self.ap.rag_runtime_service.vector_delete(collection_id, file_ids, filters)
+                return handler.ActionResponse.success(data={'count': count})
+            except Exception as e:
+                return _make_rag_error_response(e, 'VectorStoreError', collection_id=collection_id)
+
+        @self.action(PluginToRuntimeAction.GET_KNOWLEDEGE_FILE_STREAM)
+        async def get_knowledge_file_stream(data: dict[str, Any]) -> handler.ActionResponse:
+            storage_path = data['storage_path']
+            try:
+                content_bytes = await self.ap.rag_runtime_service.get_file_stream(storage_path)
+                file_key = await self.send_file(content_bytes, '')
+                return handler.ActionResponse.success(data={'file_key': file_key})
+            except Exception as e:
+                return _make_rag_error_response(e, 'FileServiceError', storage_path=storage_path)
+
+        @self.action(PluginToRuntimeAction.LIST_PARSERS)
+        async def list_parsers(data: dict[str, Any]) -> handler.ActionResponse:
+            """Plugin requests host to list available parser plugins."""
+            mime_type = data.get('mime_type')
+            try:
+                parsers = await self.ap.knowledge_service.list_parsers(mime_type)
+                return handler.ActionResponse.success(data={'parsers': parsers})
+            except Exception as e:
+                return _make_rag_error_response(e, 'ParserDiscoveryError', mime_type=mime_type)
+
+        @self.action(PluginToRuntimeAction.INVOKE_PARSER)
+        async def invoke_parser(data: dict[str, Any]) -> handler.ActionResponse:
+            """Plugin requests host to invoke a parser plugin."""
+            plugin_author = data['plugin_author']
+            plugin_name = data['plugin_name']
+            storage_path = data['storage_path']
+            mime_type = data.get('mime_type', 'application/octet-stream')
+            filename = data.get('filename', '')
+            metadata = data.get('metadata', {})
+            try:
+                # Read file from storage
+                file_bytes = await self.ap.rag_runtime_service.get_file_stream(storage_path)
+                context_data = {
+                    'mime_type': mime_type,
+                    'filename': filename,
+                    'metadata': metadata,
+                }
+                result = await self.ap.plugin_connector.call_parser(
+                    f'{plugin_author}/{plugin_name}', context_data, file_bytes
+                )
+                return handler.ActionResponse.success(data=result)
+            except Exception as e:
+                return _make_rag_error_response(e, 'ParserError')
+
+        # ================= Knowledge Base Query APIs =================
+
+        @self.action(PluginToRuntimeAction.LIST_PIPELINE_KNOWLEDGE_BASES)
+        async def list_pipeline_knowledge_bases(data: dict[str, Any]) -> handler.ActionResponse:
+            """List knowledge bases configured for the current query's pipeline."""
+            query_id = data['query_id']
+
+            if query_id not in self.ap.query_pool.cached_queries:
+                return handler.ActionResponse.error(
+                    message=f'Query with query_id {query_id} not found',
+                )
+
+            query = self.ap.query_pool.cached_queries[query_id]
+
+            kb_uuids = []
+            if query.pipeline_config:
+                local_agent_config = query.pipeline_config.get('ai', {}).get('local-agent', {})
+                kb_uuids = local_agent_config.get('knowledge-bases', [])
+                # Backward compatibility
+                if not kb_uuids:
+                    old_kb_uuid = local_agent_config.get('knowledge-base', '')
+                    if old_kb_uuid and old_kb_uuid != '__none__':
+                        kb_uuids = [old_kb_uuid]
+
+            knowledge_bases = []
+            for kb_uuid in kb_uuids:
+                kb = await self.ap.rag_mgr.get_knowledge_base_by_uuid(kb_uuid)
+                if kb:
+                    knowledge_bases.append(
+                        {
+                            'uuid': kb.get_uuid(),
+                            'name': kb.get_name(),
+                            'description': kb.knowledge_base_entity.description or '',
+                        }
+                    )
+
+            return handler.ActionResponse.success(data={'knowledge_bases': knowledge_bases})
+
+        @self.action(PluginToRuntimeAction.RETRIEVE_KNOWLEDGE_BASE)
+        async def retrieve_knowledge_base(data: dict[str, Any]) -> handler.ActionResponse:
+            """Retrieve documents from a knowledge base within the pipeline's scope."""
+            query_id = data['query_id']
+            kb_id = data['kb_id']
+            query_text = data['query_text']
+            top_k = data.get('top_k', 5)
+            filters = data.get('filters', {})
+
+            if query_id not in self.ap.query_pool.cached_queries:
+                return handler.ActionResponse.error(
+                    message=f'Query with query_id {query_id} not found',
+                )
+
+            query = self.ap.query_pool.cached_queries[query_id]
+
+            # Validate kb_id is in pipeline's allowed list
+            allowed_kb_uuids = []
+            if query.pipeline_config:
+                local_agent_config = query.pipeline_config.get('ai', {}).get('local-agent', {})
+                allowed_kb_uuids = local_agent_config.get('knowledge-bases', [])
+                if not allowed_kb_uuids:
+                    old_kb_uuid = local_agent_config.get('knowledge-base', '')
+                    if old_kb_uuid and old_kb_uuid != '__none__':
+                        allowed_kb_uuids = [old_kb_uuid]
+
+            if kb_id not in allowed_kb_uuids:
+                return handler.ActionResponse.error(
+                    message=f'Knowledge base {kb_id} is not configured for this pipeline',
+                )
+
+            kb = await self.ap.rag_mgr.get_knowledge_base_by_uuid(kb_id)
+            if not kb:
+                return handler.ActionResponse.error(
+                    message=f'Knowledge base {kb_id} not found',
+                )
+
+            try:
+                session_name = f'{query.session.launcher_type.value}_{query.session.launcher_id}'
+                entries = await kb.retrieve(
+                    query_text,
+                    settings={
+                        'top_k': top_k,
+                        'filters': filters,
+                        'session_name': session_name,
+                        'bot_uuid': query.bot_uuid or '',
+                        'sender_id': str(query.sender_id),
+                    },
+                )
+                results = [entry.model_dump(mode='json') for entry in entries]
+                return handler.ActionResponse.success(data={'results': results})
+            except Exception as e:
+                return _make_rag_error_response(e, 'RetrievalError', kb_id=kb_id)
+
+        @self.action(CommonAction.PING)
+        async def ping(data: dict[str, Any]) -> handler.ActionResponse:
+            """Ping"""
+            return handler.ActionResponse.success(
+                data={
+                    'pong': 'pong',
+                },
+            )
+
     async def ping(self) -> dict[str, Any]:
         """Ping the runtime"""
         return await self.call_action(
@@ -717,26 +959,13 @@ class RuntimeConnectionHandler(handler.Handler):
         async for ret in gen:
             yield ret
 
-    # KnowledgeRetriever methods
-    async def list_knowledge_retrievers(self, include_plugins: list[str] | None = None) -> list[dict[str, Any]]:
-        """List knowledge retrievers"""
-        result = await self.call_action(
-            LangBotToRuntimeAction.LIST_KNOWLEDGE_RETRIEVERS,
-            {
-                'include_plugins': include_plugins,
-            },
-            timeout=10,
-        )
-        return result['retrievers']
-
     async def retrieve_knowledge(
         self,
         plugin_author: str,
         plugin_name: str,
         retriever_name: str,
-        instance_id: str,
         retrieval_context: dict[str, Any],
-    ) -> list[dict[str, Any]]:
+    ) -> dict[str, Any]:
         """Retrieve knowledge"""
         result = await self.call_action(
             LangBotToRuntimeAction.RETRIEVE_KNOWLEDGE,
@@ -744,22 +973,10 @@ class RuntimeConnectionHandler(handler.Handler):
                 'plugin_author': plugin_author,
                 'plugin_name': plugin_name,
                 'retriever_name': retriever_name,
-                'instance_id': instance_id,
                 'retrieval_context': retrieval_context,
             },
             timeout=30,
         )
-        return result['retrieval_results']
-
-    async def sync_polymorphic_component_instances(self, required_instances: list[dict[str, Any]]) -> dict[str, Any]:
-        """Sync polymorphic component instances with runtime"""
-        result = await self.call_action(
-            LangBotToRuntimeAction.SYNC_POLYMORPHIC_COMPONENT_INSTANCES,
-            {
-                'required_instances': required_instances,
-            },
-            timeout=30,
-        )
         return result
 
     async def get_debug_info(self) -> dict[str, Any]:
@@ -770,3 +987,91 @@ class RuntimeConnectionHandler(handler.Handler):
             timeout=10,
         )
         return result
+
+    # ================= RAG Capability Callers (LangBot -> Runtime) =================
+
+    async def rag_ingest_document(
+        self, plugin_author: str, plugin_name: str, context_data: dict[str, Any]
+    ) -> dict[str, Any]:
+        """Send INGEST_DOCUMENT action to runtime."""
+        result = await self.call_action(
+            LangBotToRuntimeAction.RAG_INGEST_DOCUMENT,
+            {'plugin_author': plugin_author, 'plugin_name': plugin_name, 'context': context_data},
+            timeout=1200,  # Ingestion can be slow for large documents
+        )
+        return result
+
+    async def rag_delete_document(self, plugin_author: str, plugin_name: str, document_id: str, kb_id: str) -> bool:
+        result = await self.call_action(
+            LangBotToRuntimeAction.RAG_DELETE_DOCUMENT,
+            {'plugin_author': plugin_author, 'plugin_name': plugin_name, 'document_id': document_id, 'kb_id': kb_id},
+            timeout=30,
+        )
+        return result.get('success', False)
+
+    async def rag_on_kb_create(
+        self, plugin_author: str, plugin_name: str, kb_id: str, config: dict[str, Any]
+    ) -> dict[str, Any]:
+        """Notify plugin about KB creation."""
+        result = await self.call_action(
+            LangBotToRuntimeAction.RAG_ON_KB_CREATE,
+            {'plugin_author': plugin_author, 'plugin_name': plugin_name, 'kb_id': kb_id, 'config': config},
+            timeout=30,
+        )
+        return result
+
+    async def rag_on_kb_delete(self, plugin_author: str, plugin_name: str, kb_id: str) -> dict[str, Any]:
+        """Notify plugin about KB deletion."""
+        result = await self.call_action(
+            LangBotToRuntimeAction.RAG_ON_KB_DELETE,
+            {'plugin_author': plugin_author, 'plugin_name': plugin_name, 'kb_id': kb_id},
+            timeout=30,
+        )
+        return result
+
+    async def get_rag_creation_schema(self, plugin_author: str, plugin_name: str) -> dict[str, Any]:
+        return await self.call_action(
+            LangBotToRuntimeAction.GET_RAG_CREATION_SETTINGS_SCHEMA,
+            {'plugin_author': plugin_author, 'plugin_name': plugin_name},
+            timeout=10,
+        )
+
+    async def get_rag_retrieval_schema(self, plugin_author: str, plugin_name: str) -> dict[str, Any]:
+        return await self.call_action(
+            LangBotToRuntimeAction.GET_RAG_RETRIEVAL_SETTINGS_SCHEMA,
+            {'plugin_author': plugin_author, 'plugin_name': plugin_name},
+            timeout=10,
+        )
+
+    async def list_knowledge_engines(self) -> list[dict[str, Any]]:
+        """List all available Knowledge Engines from plugins."""
+        result = await self.call_action(LangBotToRuntimeAction.LIST_KNOWLEDGE_ENGINES, {}, timeout=60)
+        return result.get('engines', [])
+
+    # ================= Parser Capability Callers (LangBot -> Runtime) =================
+
+    async def list_parsers(self) -> list[dict[str, Any]]:
+        """List all available parsers from plugins."""
+        result = await self.call_action(LangBotToRuntimeAction.LIST_PARSERS, {}, timeout=60)
+        return result.get('parsers', [])
+
+    async def parse_document(
+        self, plugin_author: str, plugin_name: str, context_data: dict[str, Any], file_bytes: bytes
+    ) -> dict[str, Any]:
+        """Send PARSE_DOCUMENT action to runtime.
+
+        Sends file content via chunked FILE_CHUNK transfer, then invokes
+        the PARSE_DOCUMENT action with a file_key reference.
+        """
+        # Send file to runtime via chunked transfer
+        file_key = await self.send_file(file_bytes, '')
+
+        # Include file_key in context_data for the runtime to read
+        context_data['file_key'] = file_key
+
+        result = await self.call_action(
+            LangBotToRuntimeAction.PARSE_DOCUMENT,
+            {'plugin_author': plugin_author, 'plugin_name': plugin_name, 'context': context_data},
+            timeout=300,
+        )
+        return result

src/langbot/pkg/provider/runners/difysvapi.py

@@ -72,6 +72,28 @@ class DifyServiceAPIRunner(runner.RequestRunner):
                 content = f'<think>\n{thinking_content}\n</think>\n{content}'.strip()
             return content, thinking_content
 
+    def _extract_dify_text_output(self, value: typing.Any) -> str:
+        """Extract text content from Dify output payload."""
+        if value is None:
+            return ''
+        if isinstance(value, dict):
+            content = value.get('content')
+            if isinstance(content, str):
+                return content
+            return json.dumps(value, ensure_ascii=False)
+        if isinstance(value, str):
+            text = value.strip()
+            if not text:
+                return ''
+            try:
+                parsed = json.loads(text)
+            except json.JSONDecodeError:
+                return value
+            if isinstance(parsed, dict) and isinstance(parsed.get('content'), str):
+                return parsed['content']
+            return value
+        return str(value)
+
     async def _preprocess_user_message(self, query: pipeline_query.Query) -> tuple[str, list[dict]]:
         """预处理用户消息，提取纯文本，并将图片/文件上传到 Dify 服务
 
@@ -192,7 +214,8 @@ class DifyServiceAPIRunner(runner.RequestRunner):
             if mode == 'workflow':
                 if chunk['event'] == 'node_finished':
                     if chunk['data']['node_type'] == 'answer':
-                        content, _ = self._process_thinking_content(chunk['data']['outputs']['answer'])
+                        answer = self._extract_dify_text_output(chunk['data']['outputs'].get('answer'))
+                        content, _ = self._process_thinking_content(answer)
 
                         yield provider_message.Message(
                             role='assistant',
@@ -405,6 +428,7 @@ class DifyServiceAPIRunner(runner.RequestRunner):
             for f in upload_files
         ]
 
+        mode = 'basic'
         basic_mode_pending_chunk = ''
 
         inputs = {}
@@ -417,6 +441,7 @@ class DifyServiceAPIRunner(runner.RequestRunner):
         is_final = False
         think_start = False
         think_end = False
+        yielded_final = False
 
         remove_think = self.pipeline_config['output'].get('misc', '').get('remove-think')
 
@@ -430,11 +455,12 @@ class DifyServiceAPIRunner(runner.RequestRunner):
         ):
             self.ap.logger.debug('dify-chat-chunk: ' + str(chunk))
 
-            # if chunk['event'] == 'workflow_started':
-            #     mode = 'workflow'
-            # if mode == 'workflow':
-            # elif mode == 'basic':
-            # 因为都只是返回的 message也没有工具调用什么的，暂时不分类
+            if chunk['event'] == 'workflow_started':
+                mode = 'workflow'
+            elif chunk['event'] in ('node_started', 'node_finished', 'workflow_finished'):
+                # Some Dify deployments may omit workflow_started in streamed chunks.
+                mode = 'workflow'
+
             if chunk['event'] == 'message':
                 message_idx += 1
                 if remove_think:
@@ -457,14 +483,30 @@ class DifyServiceAPIRunner(runner.RequestRunner):
 
             if chunk['event'] == 'message_end':
                 is_final = True
+            elif chunk['event'] == 'workflow_finished':
+                is_final = True
+                if chunk['data'].get('error'):
+                    raise errors.DifyAPIError(chunk['data']['error'])
 
-            if is_final or message_idx % 8 == 0:
+            if mode == 'workflow' and chunk['event'] == 'node_finished':
+                if chunk['data'].get('node_type') == 'answer':
+                    answer = self._extract_dify_text_output(chunk['data'].get('outputs', {}).get('answer'))
+                    if answer:
+                        basic_mode_pending_chunk = answer
+
+            if (
+                not yielded_final
+                and (is_final or message_idx % 8 == 0)
+                and (basic_mode_pending_chunk != '' or is_final)
+            ):
                 # content, _ = self._process_thinking_content(basic_mode_pending_chunk)
                 yield provider_message.MessageChunk(
                     role='assistant',
                     content=basic_mode_pending_chunk,
                     is_final=is_final,
                 )
+                if is_final:
+                    yielded_final = True
 
         if chunk is None:
             raise errors.DifyAPIError('Dify API 没有返回任何响应，请检查网络连接和API配置')

src/langbot/pkg/provider/runners/localagent.py

@@ -4,6 +4,7 @@ import json
 import copy
 import typing
 from .. import runner
+from ..modelmgr import requester as modelmgr_requester
 import langbot_plugin.api.entities.builtin.pipeline.query as pipeline_query
 import langbot_plugin.api.entities.builtin.provider.message as provider_message
 import langbot_plugin.api.entities.builtin.rag.context as rag_context
@@ -26,29 +27,114 @@ Respond in the same language as the user's input.
 
 @runner.runner_class('local-agent')
 class LocalAgentRunner(runner.RequestRunner):
-    """本地Agent请求运行器"""
+    """Local agent request runner"""
 
-    class ToolCallTracker:
-        """工具调用追踪器"""
+    async def _get_model_candidates(
+        self,
+        query: pipeline_query.Query,
+    ) -> list[modelmgr_requester.RuntimeLLMModel]:
+        """Build ordered list of models to try: primary model + fallback models."""
+        candidates = []
 
-        def __init__(self):
-            self.active_calls: dict[str, dict] = {}
-            self.completed_calls: list[provider_message.ToolCall] = []
+        # Primary model
+        if query.use_llm_model_uuid:
+            try:
+                primary = await self.ap.model_mgr.get_model_by_uuid(query.use_llm_model_uuid)
+                candidates.append(primary)
+            except ValueError:
+                self.ap.logger.warning(f'Primary model {query.use_llm_model_uuid} not found')
+
+        # Fallback models
+        fallback_uuids = (query.variables or {}).get('_fallback_model_uuids', [])
+        for fb_uuid in fallback_uuids:
+            try:
+                fb_model = await self.ap.model_mgr.get_model_by_uuid(fb_uuid)
+                candidates.append(fb_model)
+            except ValueError:
+                self.ap.logger.warning(f'Fallback model {fb_uuid} not found, skipping')
+
+        return candidates
+
+    async def _invoke_with_fallback(
+        self,
+        query: pipeline_query.Query,
+        candidates: list[modelmgr_requester.RuntimeLLMModel],
+        messages: list,
+        funcs: list,
+        remove_think: bool,
+    ) -> tuple[provider_message.Message, modelmgr_requester.RuntimeLLMModel]:
+        """Try non-streaming invocation with sequential fallback. Returns (message, model_used)."""
+        last_error = None
+        for model in candidates:
+            try:
+                msg = await model.provider.invoke_llm(
+                    query,
+                    model,
+                    messages,
+                    funcs if model.model_entity.abilities.__contains__('func_call') else [],
+                    extra_args=model.model_entity.extra_args,
+                    remove_think=remove_think,
+                )
+                return msg, model
+            except Exception as e:
+                last_error = e
+                self.ap.logger.warning(f'Model {model.model_entity.name} failed: {e}, trying next fallback...')
+        raise last_error or RuntimeError('No model candidates available')
+
+    async def _invoke_stream_with_fallback(
+        self,
+        query: pipeline_query.Query,
+        candidates: list[modelmgr_requester.RuntimeLLMModel],
+        messages: list,
+        funcs: list,
+        remove_think: bool,
+    ) -> tuple[typing.AsyncGenerator, modelmgr_requester.RuntimeLLMModel]:
+        """Try streaming invocation with sequential fallback. Returns (stream_generator, model_used).
+
+        Fallback is only possible before any chunks have been yielded to the client.
+        Once streaming starts, the model is committed.
+        """
+        last_error = None
+        for model in candidates:
+            try:
+                stream = model.provider.invoke_llm_stream(
+                    query,
+                    model,
+                    messages,
+                    funcs if model.model_entity.abilities.__contains__('func_call') else [],
+                    extra_args=model.model_entity.extra_args,
+                    remove_think=remove_think,
+                )
+                # Attempt to get the first chunk to verify the stream works
+                first_chunk = await stream.__anext__()
+
+                async def _chain_stream(first, rest):
+                    yield first
+                    async for chunk in rest:
+                        yield chunk
+
+                return _chain_stream(first_chunk, stream), model
+            except StopAsyncIteration:
+                # Empty stream — treat as success (model returned nothing)
+                async def _empty_stream():
+                    return
+                    yield  # make it a generator
+
+                return _empty_stream(), model
+            except Exception as e:
+                last_error = e
+                self.ap.logger.warning(f'Model {model.model_entity.name} stream failed: {e}, trying next fallback...')
+        raise last_error or RuntimeError('No model candidates available')
 
     async def run(
         self, query: pipeline_query.Query
     ) -> typing.AsyncGenerator[provider_message.Message | provider_message.MessageChunk, None]:
-        """运行请求"""
+        """Run request"""
         pending_tool_calls = []
 
-        # Get knowledge bases list (new field)
-        kb_uuids = query.pipeline_config['ai']['local-agent'].get('knowledge-bases', [])
-
-        # Fallback to old field for backward compatibility
-        if not kb_uuids:
-            old_kb_uuid = query.pipeline_config['ai']['local-agent'].get('knowledge-base', '')
-            if old_kb_uuid and old_kb_uuid != '__none__':
-                kb_uuids = [old_kb_uuid]
+        # Get knowledge bases list from query variables (set by PreProcessor,
+        # may have been modified by plugins during PromptPreProcessing)
+        kb_uuids = query.variables.get('_knowledge_base_uuids', [])
 
         user_message = copy.deepcopy(query.user_message)
 
@@ -74,15 +160,14 @@ class LocalAgentRunner(runner.RequestRunner):
                     self.ap.logger.warning(f'Knowledge base {kb_uuid} not found, skipping')
                     continue
 
-                # Get top_k based on KB type
-                if kb.get_type() == 'internal':
-                    top_k = kb.knowledge_base_entity.top_k
-                elif kb.get_type() == 'external':
-                    top_k = 5  # external kb's top_k is managed by plugin config
-                else:
-                    top_k = 5  # default fallback
-
-                result = await kb.retrieve(user_message_text, top_k)
+                result = await kb.retrieve(
+                    user_message_text,
+                    settings={
+                        'bot_uuid': query.bot_uuid or '',
+                        'sender_id': str(query.sender_id),
+                        'session_name': f'{query.session.launcher_type.value}_{query.session.launcher_id}',
+                    },
+                )
 
                 if result:
                     all_results.extend(result)
@@ -97,9 +182,9 @@ class LocalAgentRunner(runner.RequestRunner):
                         if content.type == 'text' and content.text is not None:
                             texts.append(f'[{idx}] {content.text}')
                             idx += 1
-                rag_context = '\n\n'.join(texts)
+                rag_context_text = '\n\n'.join(texts)
                 final_user_message_text = rag_combined_prompt_template.format(
-                    rag_context=rag_context, user_message=user_message_text
+                    rag_context=rag_context_text, user_message=user_message_text
                 )
 
             else:
@@ -121,51 +206,51 @@ class LocalAgentRunner(runner.RequestRunner):
 
         remove_think = query.pipeline_config['output'].get('misc', '').get('remove-think')
 
-        use_llm_model = await self.ap.model_mgr.get_model_by_uuid(query.use_llm_model_uuid)
+        # Build ordered candidate list (primary + fallbacks)
+        candidates = await self._get_model_candidates(query)
+        if not candidates:
+            raise RuntimeError('No LLM model configured for local-agent runner')
 
         self.ap.logger.debug(
-            f'localagent req: query={query.query_id} req_messages={req_messages} use_llm_model={query.use_llm_model_uuid}'
+            f'localagent req: query={query.query_id} req_messages={req_messages} '
+            f'candidates={[m.model_entity.name for m in candidates]}'
         )
 
         if not is_stream:
-            # 非流式输出，直接请求
-
-            msg = await use_llm_model.provider.invoke_llm(
+            # Non-streaming: invoke with fallback
+            msg, use_llm_model = await self._invoke_with_fallback(
                 query,
-                use_llm_model,
+                candidates,
                 req_messages,
                 query.use_funcs,
-                extra_args=use_llm_model.model_entity.extra_args,
-                remove_think=remove_think,
+                remove_think,
             )
             yield msg
             final_msg = msg
         else:
-            # 流式输出，需要处理工具调用
+            # Streaming: invoke with fallback
             tool_calls_map: dict[str, provider_message.ToolCall] = {}
             msg_idx = 0
-            accumulated_content = ''  # 从开始累积的所有内容
+            accumulated_content = ''
             last_role = 'assistant'
             msg_sequence = 1
-            async for msg in use_llm_model.provider.invoke_llm_stream(
+
+            stream_src, use_llm_model = await self._invoke_stream_with_fallback(
                 query,
-                use_llm_model,
+                candidates,
                 req_messages,
                 query.use_funcs,
-                extra_args=use_llm_model.model_entity.extra_args,
-                remove_think=remove_think,
-            ):
+                remove_think,
+            )
+            async for msg in stream_src:
                 msg_idx = msg_idx + 1
 
-                # 记录角色
                 if msg.role:
                     last_role = msg.role
 
-                # 累积内容
                 if msg.content:
                     accumulated_content += msg.content
 
-                # 处理工具调用
                 if msg.tool_calls:
                     for tool_call in msg.tool_calls:
                         if tool_call.id not in tool_calls_map:
@@ -177,21 +262,18 @@ class LocalAgentRunner(runner.RequestRunner):
                                 ),
                             )
                         if tool_call.function and tool_call.function.arguments:
-                            # 流式处理中，工具调用参数可能分多个chunk返回，需要追加而不是覆盖
                             tool_calls_map[tool_call.id].function.arguments += tool_call.function.arguments
-                # continue
-                # 每8个chunk或最后一个chunk时，输出所有累积的内容
+
                 if msg_idx % 8 == 0 or msg.is_final:
                     msg_sequence += 1
                     yield provider_message.MessageChunk(
                         role=last_role,
-                        content=accumulated_content,  # 输出所有累积内容
+                        content=accumulated_content,
                         tool_calls=list(tool_calls_map.values()) if (tool_calls_map and msg.is_final) else None,
                         is_final=msg.is_final,
                         msg_sequence=msg_sequence,
                     )
 
-            # 创建最终消息用于后续处理
             final_msg = provider_message.MessageChunk(
                 role=last_role,
                 content=accumulated_content,
@@ -206,7 +288,8 @@ class LocalAgentRunner(runner.RequestRunner):
 
         req_messages.append(final_msg)
 
-        # 持续请求，只要还有待处理的工具调用就继续处理调用
+        # Once a model succeeds, commit to it for the tool call loop
+        # (no fallback mid-conversation — different models may interpret tool results differently)
         while pending_tool_calls:
             for tool_call in pending_tool_calls:
                 try:
@@ -247,7 +330,6 @@ class LocalAgentRunner(runner.RequestRunner):
 
                     req_messages.append(msg)
                 except Exception as e:
-                    # 工具调用出错，添加一个报错信息到 req_messages
                     err_msg = provider_message.Message(role='tool', content=f'err: {e}', tool_call_id=tool_call.id)
 
                     yield err_msg
@@ -255,39 +337,38 @@ class LocalAgentRunner(runner.RequestRunner):
                     req_messages.append(err_msg)
 
             self.ap.logger.debug(
-                f'localagent req: query={query.query_id} req_messages={req_messages} use_llm_model={query.use_llm_model_uuid}'
+                f'localagent req: query={query.query_id} req_messages={req_messages} '
+                f'use_llm_model={use_llm_model.model_entity.name}'
             )
 
             if is_stream:
                 tool_calls_map = {}
                 msg_idx = 0
-                accumulated_content = ''  # 从开始累积的所有内容
+                accumulated_content = ''
                 last_role = 'assistant'
                 msg_sequence = first_end_sequence
 
-                async for msg in use_llm_model.provider.invoke_llm_stream(
+                tool_stream_src = use_llm_model.provider.invoke_llm_stream(
                     query,
                     use_llm_model,
                     req_messages,
-                    query.use_funcs,
+                    query.use_funcs if use_llm_model.model_entity.abilities.__contains__('func_call') else [],
                     extra_args=use_llm_model.model_entity.extra_args,
                     remove_think=remove_think,
-                ):
+                )
+                async for msg in tool_stream_src:
                     msg_idx += 1
 
-                    # 记录角色
                     if msg.role:
                         last_role = msg.role
 
-                    # 第一次请求工具调用时的内容
+                    # Prepend first-round content on first chunk of tool-call round
                     if msg_idx == 1:
                         accumulated_content = first_content if first_content is not None else accumulated_content
 
-                    # 累积内容
                     if msg.content:
                         accumulated_content += msg.content
 
-                    # 处理工具调用
                     if msg.tool_calls:
                         for tool_call in msg.tool_calls:
                             if tool_call.id not in tool_calls_map:
@@ -299,15 +380,13 @@ class LocalAgentRunner(runner.RequestRunner):
                                     ),
                                 )
                             if tool_call.function and tool_call.function.arguments:
-                                # 流式处理中，工具调用参数可能分多个chunk返回，需要追加而不是覆盖
                                 tool_calls_map[tool_call.id].function.arguments += tool_call.function.arguments
 
-                    # 每8个chunk或最后一个chunk时，输出所有累积的内容
                     if msg_idx % 8 == 0 or msg.is_final:
                         msg_sequence += 1
                         yield provider_message.MessageChunk(
                             role=last_role,
-                            content=accumulated_content,  # 输出所有累积内容
+                            content=accumulated_content,
                             tool_calls=list(tool_calls_map.values()) if (tool_calls_map and msg.is_final) else None,
                             is_final=msg.is_final,
                             msg_sequence=msg_sequence,
@@ -320,12 +399,12 @@ class LocalAgentRunner(runner.RequestRunner):
                     msg_sequence=msg_sequence,
                 )
             else:
-                # 处理完所有调用，再次请求
+                # Non-streaming: use committed model directly (no fallback in tool loop)
                 msg = await use_llm_model.provider.invoke_llm(
                     query,
                     use_llm_model,
                     req_messages,
-                    query.use_funcs,
+                    query.use_funcs if use_llm_model.model_entity.abilities.__contains__('func_call') else [],
                     extra_args=use_llm_model.model_entity.extra_args,
                     remove_think=remove_think,
                 )

src/langbot/pkg/provider/runners/n8nsvapi.py

@@ -5,6 +5,8 @@ import json
 import uuid
 import aiohttp
 
+from langbot.pkg.utils import httpclient
+
 from .. import runner
 from ...core import app
 import langbot_plugin.api.entities.builtin.pipeline.query as pipeline_query
@@ -217,50 +219,50 @@ class N8nServiceAPIRunner(runner.RequestRunner):
                 self.ap.logger.debug('no auth')
 
             # 调用webhook
-            async with aiohttp.ClientSession() as session:
-                if is_stream:
-                    # 流式请求
-                    async with session.post(
-                        self.webhook_url, json=payload, headers=headers, auth=auth, timeout=self.timeout
-                    ) as response:
+            session = httpclient.get_session()
+            if is_stream:
+                # 流式请求
+                async with session.post(
+                    self.webhook_url, json=payload, headers=headers, auth=auth, timeout=self.timeout
+                ) as response:
+                    if response.status != 200:
+                        error_text = await response.text()
+                        self.ap.logger.error(f'n8n webhook call failed: {response.status}, {error_text}')
+                        raise Exception(f'n8n webhook call failed: {response.status}, {error_text}')
+
+                    # 处理流式响应
+                    async for chunk in self._process_stream_response(response):
+                        yield chunk
+            else:
+                async with session.post(
+                    self.webhook_url, json=payload, headers=headers, auth=auth, timeout=self.timeout
+                ) as response:
+                    try:
+                        async for chunk in self._process_stream_response(response):
+                            output_content = chunk.content if chunk.is_final else ''
+                    except:
+                        # 非流式请求（保持原有逻辑）
                         if response.status != 200:
                             error_text = await response.text()
                             self.ap.logger.error(f'n8n webhook call failed: {response.status}, {error_text}')
                             raise Exception(f'n8n webhook call failed: {response.status}, {error_text}')
 
-                        # 处理流式响应
-                        async for chunk in self._process_stream_response(response):
-                            yield chunk
-                else:
-                    async with session.post(
-                        self.webhook_url, json=payload, headers=headers, auth=auth, timeout=self.timeout
-                    ) as response:
-                        try:
-                            async for chunk in self._process_stream_response(response):
-                                output_content = chunk.content if chunk.is_final else ''
-                        except:
-                            # 非流式请求（保持原有逻辑）
-                            if response.status != 200:
-                                error_text = await response.text()
-                                self.ap.logger.error(f'n8n webhook call failed: {response.status}, {error_text}')
-                                raise Exception(f'n8n webhook call failed: {response.status}, {error_text}')
+                        # 解析响应
+                        response_data = await response.json()
+                        self.ap.logger.debug(f'n8n webhook response: {response_data}')
 
-                            # 解析响应
-                            response_data = await response.json()
-                            self.ap.logger.debug(f'n8n webhook response: {response_data}')
+                        # 从响应中提取输出
+                        if self.output_key in response_data:
+                            output_content = response_data[self.output_key]
+                        else:
+                            # 如果没有指定的输出键，则使用整个响应
+                            output_content = json.dumps(response_data, ensure_ascii=False)
 
-                            # 从响应中提取输出
-                            if self.output_key in response_data:
-                                output_content = response_data[self.output_key]
-                            else:
-                                # 如果没有指定的输出键，则使用整个响应
-                                output_content = json.dumps(response_data, ensure_ascii=False)
-
-                        # 返回消息
-                        yield provider_message.Message(
-                            role='assistant',
-                            content=output_content,
-                        )
+                    # 返回消息
+                    yield provider_message.Message(
+                        role='assistant',
+                        content=output_content,
+                    )
         except Exception as e:
             self.ap.logger.error(f'n8n webhook call exception: {str(e)}')
             raise N8nAPIError(f'n8n webhook call exception: {str(e)}')

src/langbot/pkg/rag/knowledge/base.py

@@ -22,12 +22,12 @@ class KnowledgeBaseInterface(metaclass=abc.ABCMeta):
         pass
 
     @abc.abstractmethod
-    async def retrieve(self, query: str, top_k: int) -> list[rag_context.RetrievalResultEntry]:
+    async def retrieve(self, query: str, settings: dict | None = None) -> list[rag_context.RetrievalResultEntry]:
         """Retrieve relevant documents from the knowledge base
 
         Args:
             query: The query string
-            top_k: Number of top results to return
+            settings: Optional per-request retrieval settings overrides
 
         Returns:
             List of retrieve result entries
@@ -45,8 +45,8 @@ class KnowledgeBaseInterface(metaclass=abc.ABCMeta):
         pass
 
     @abc.abstractmethod
-    def get_type(self) -> str:
-        """Get the type of knowledge base (internal/external)"""
+    def get_knowledge_engine_plugin_id(self) -> str:
+        """Get the Knowledge Engine plugin ID"""
         pass
 
     @abc.abstractmethod

src/langbot/pkg/rag/knowledge/external.py

@@ -1,85 +0,0 @@
-"""External knowledge base implementation"""
-
-from __future__ import annotations
-
-from langbot.pkg.core import app
-from langbot.pkg.entity.persistence import rag as persistence_rag
-from langbot_plugin.api.entities.builtin.rag import context as rag_context
-from .base import KnowledgeBaseInterface
-
-
-class ExternalKnowledgeBase(KnowledgeBaseInterface):
-    """External knowledge base that queries via HTTP API or plugin retriever"""
-
-    external_kb_entity: persistence_rag.ExternalKnowledgeBase
-
-    # Plugin retriever instance ID
-    retriever_instance_id: str | None
-
-    def __init__(self, ap: app.Application, external_kb_entity: persistence_rag.ExternalKnowledgeBase):
-        super().__init__(ap)
-        self.external_kb_entity = external_kb_entity
-        self.retriever_instance_id = None
-
-    async def initialize(self):
-        """Initialize the external knowledge base"""
-        # Use KB UUID as instance ID
-        # Instance creation is now handled by the unified sync mechanism
-        # when LangBot connects to runtime
-        self.retriever_instance_id = self.external_kb_entity.uuid
-
-        self.ap.logger.info(
-            f'Initialized external KB {self.external_kb_entity.uuid}, instance will be created by sync mechanism'
-        )
-
-    async def retrieve(self, query: str, top_k: int = 5) -> list[rag_context.RetrievalResultEntry]:
-        """Retrieve documents from external knowledge base via plugin retriever"""
-        if not self.retriever_instance_id:
-            self.ap.logger.error(f'No retriever instance for KB {self.external_kb_entity.uuid}')
-            return []
-
-        try:
-            results = await self.ap.plugin_connector.retrieve_knowledge(
-                self.external_kb_entity.plugin_author,
-                self.external_kb_entity.plugin_name,
-                self.external_kb_entity.retriever_name,
-                self.retriever_instance_id,
-                {'query': query},
-            )
-
-            # Convert plugin results to RetrievalResultEntry
-            retrieval_entries = []
-            for result in results:
-                retrieval_entries.append(rag_context.RetrievalResultEntry(**result))
-
-            return retrieval_entries
-        except Exception as e:
-            self.ap.logger.error(f'Plugin retriever error: {e}')
-            import traceback
-
-            traceback.print_exc()
-            return []
-
-    def get_uuid(self) -> str:
-        """Get the UUID of the external knowledge base"""
-        return self.external_kb_entity.uuid
-
-    def get_name(self) -> str:
-        """Get the name of the external knowledge base"""
-        return self.external_kb_entity.name
-
-    def get_type(self) -> str:
-        """Get the type of knowledge base"""
-        return 'external'
-
-    async def dispose(self):
-        """Clean up resources"""
-        # Trigger sync to immediately delete the instance from plugin process
-        # This ensures instance is cleaned up without waiting for next LangBot restart
-        try:
-            await self.ap.plugin_connector.sync_polymorphic_component_instances()
-            self.ap.logger.info(
-                f'Disposed external KB {self.external_kb_entity.uuid}, triggered sync to delete instance'
-            )
-        except Exception as e:
-            self.ap.logger.error(f'Failed to sync after disposing KB: {e}')

src/langbot/pkg/rag/knowledge/kbmgr.py

@@ -1,18 +1,19 @@
 from __future__ import annotations
+import mimetypes
+import os.path
 import traceback
 import uuid
 import zipfile
 import io
-from .services import parser, chunker
+from typing import Any
 from langbot.pkg.core import app
-from langbot.pkg.rag.knowledge.services.embedder import Embedder
-from langbot.pkg.rag.knowledge.services.retriever import Retriever
 import sqlalchemy
+
+
 from langbot.pkg.entity.persistence import rag as persistence_rag
 from langbot.pkg.core import taskmgr
 from langbot_plugin.api.entities.builtin.rag import context as rag_context
 from .base import KnowledgeBaseInterface
-from .external import ExternalKnowledgeBase
 
 
 class RuntimeKnowledgeBase(KnowledgeBaseInterface):
@@ -20,28 +21,16 @@ class RuntimeKnowledgeBase(KnowledgeBaseInterface):
 
     knowledge_base_entity: persistence_rag.KnowledgeBase
 
-    parser: parser.FileParser
-
-    chunker: chunker.Chunker
-
-    embedder: Embedder
-
-    retriever: Retriever
-
     def __init__(self, ap: app.Application, knowledge_base_entity: persistence_rag.KnowledgeBase):
         super().__init__(ap)
         self.knowledge_base_entity = knowledge_base_entity
-        self.parser = parser.FileParser(ap=self.ap)
-        self.chunker = chunker.Chunker(ap=self.ap)
-        self.embedder = Embedder(ap=self.ap)
-        self.retriever = Retriever(ap=self.ap)
-        # 传递kb_id给retriever
-        self.retriever.kb_id = knowledge_base_entity.uuid
 
     async def initialize(self):
         pass
 
-    async def _store_file_task(self, file: persistence_rag.File, task_context: taskmgr.TaskContext):
+    async def _store_file_task(
+        self, file: persistence_rag.File, task_context: taskmgr.TaskContext, parser_plugin_id: str | None = None
+    ):
         try:
             # set file status to processing
             await self.ap.persistence_mgr.execute_async(
@@ -50,31 +39,46 @@ class RuntimeKnowledgeBase(KnowledgeBaseInterface):
                 .values(status='processing')
             )
 
-            task_context.set_current_action('Parsing file')
-            # parse file
-            text = await self.parser.parse(file.file_name, file.extension)
-            if not text:
-                raise Exception(f'No text extracted from file {file.file_name}')
+            task_context.set_current_action('Processing file')
 
-            task_context.set_current_action('Chunking file')
-            # chunk file
-            chunks_texts = await self.chunker.chunk(text)
-            if not chunks_texts:
-                raise Exception(f'No chunks extracted from file {file.file_name}')
+            # Get file size from storage
+            file_size = await self.ap.storage_mgr.storage_provider.size(file.file_name)
 
-            task_context.set_current_action('Embedding chunks')
+            # Detect MIME type from extension
+            mime_type, _ = mimetypes.guess_type(file.file_name)
+            if mime_type is None:
+                mime_type = 'application/octet-stream'
 
-            embedding_model = await self.ap.model_mgr.get_embedding_model_by_uuid(
-                self.knowledge_base_entity.embedding_model_uuid
-            )
-            # embed chunks
-            await self.embedder.embed_and_store(
-                kb_id=self.knowledge_base_entity.uuid,
-                file_id=file.uuid,
-                chunks=chunks_texts,
-                embedding_model=embedding_model,
+            # If a parser plugin is specified, call it before ingestion
+            parsed_content = None
+            if parser_plugin_id:
+                task_context.set_current_action('Parsing file')
+                file_bytes = await self.ap.storage_mgr.storage_provider.load(file.file_name)
+                parse_context = {
+                    'mime_type': mime_type,
+                    'filename': file.file_name,
+                    'metadata': {},
+                }
+                parsed_content = await self.ap.plugin_connector.call_parser(parser_plugin_id, parse_context, file_bytes)
+
+            # Call plugin to ingest document
+            result = await self._ingest_document(
+                {
+                    'document_id': file.uuid,
+                    'filename': file.file_name,
+                    'extension': file.extension,
+                    'file_size': file_size,
+                    'mime_type': mime_type,
+                },
+                file.file_name,  # storage path
+                parsed_content=parsed_content,
             )
 
+            # Check plugin result status
+            if result.get('status') == 'failed':
+                error_msg = result.get('error_message', 'Plugin ingestion returned failed status')
+                raise Exception(error_msg)
+
             # set file status to completed
             await self.ap.persistence_mgr.execute_async(
                 sqlalchemy.update(persistence_rag.File)
@@ -97,16 +101,17 @@ class RuntimeKnowledgeBase(KnowledgeBaseInterface):
             # delete file from storage
             await self.ap.storage_mgr.storage_provider.delete(file.file_name)
 
-    async def store_file(self, file_id: str) -> str:
+    async def store_file(self, file_id: str, parser_plugin_id: str | None = None) -> str:
         # pre checking
         if not await self.ap.storage_mgr.storage_provider.exists(file_id):
             raise Exception(f'File {file_id} not found')
 
         file_name = file_id
-        extension = file_name.split('.')[-1].lower()
+        _, ext = os.path.splitext(file_name)
+        extension = ext.lstrip('.').lower() if ext else ''
 
         if extension == 'zip':
-            return await self._store_zip_file(file_id)
+            return await self._store_zip_file(file_id, parser_plugin_id=parser_plugin_id)
 
         file_uuid = str(uuid.uuid4())
         kb_id = self.knowledge_base_entity.uuid
@@ -126,7 +131,7 @@ class RuntimeKnowledgeBase(KnowledgeBaseInterface):
         # run background task asynchronously
         ctx = taskmgr.TaskContext.new()
         wrapper = self.ap.task_mgr.create_user_task(
-            self._store_file_task(file_obj, task_context=ctx),
+            self._store_file_task(file_obj, task_context=ctx, parser_plugin_id=parser_plugin_id),
             kind='knowledge-operation',
             name=f'knowledge-store-file-{file_id}',
             label=f'Store file {file_id}',
@@ -134,7 +139,7 @@ class RuntimeKnowledgeBase(KnowledgeBaseInterface):
         )
         return wrapper.id
 
-    async def _store_zip_file(self, zip_file_id: str) -> str:
+    async def _store_zip_file(self, zip_file_id: str, parser_plugin_id: str | None = None) -> str:
         """Handle ZIP file by extracting each document and storing them separately."""
         self.ap.logger.info(f'Processing ZIP file: {zip_file_id}')
 
@@ -150,7 +155,8 @@ class RuntimeKnowledgeBase(KnowledgeBaseInterface):
                 if file_info.is_dir() or file_info.filename.startswith('.'):
                     continue
 
-                file_extension = file_info.filename.split('.')[-1].lower()
+                _, file_ext = os.path.splitext(file_info.filename)
+                file_extension = file_ext.lstrip('.').lower()
                 if file_extension not in supported_extensions:
                     self.ap.logger.debug(f'Skipping unsupported file in ZIP: {file_info.filename}')
                     continue
@@ -159,18 +165,18 @@ class RuntimeKnowledgeBase(KnowledgeBaseInterface):
                     file_content = zip_ref.read(file_info.filename)
 
                     base_name = file_info.filename.replace('/', '_').replace('\\', '_')
-                    extension = base_name.split('.')[-1]
-                    file_name = base_name.split('.')[0]
+                    file_stem, file_ext = os.path.splitext(base_name)
+                    extension = file_ext.lstrip('.')
 
-                    if file_name.startswith('__MACOSX'):
+                    if file_stem.startswith('__MACOSX'):
                         continue
 
-                    extracted_file_id = file_name + '_' + str(uuid.uuid4())[:8] + '.' + extension
+                    extracted_file_id = file_stem + '_' + str(uuid.uuid4())[:8] + '.' + extension
                     # save file to storage
 
                     await self.ap.storage_mgr.storage_provider.save(extracted_file_id, file_content)
 
-                    task_id = await self.store_file(extracted_file_id)
+                    task_id = await self.store_file(extracted_file_id, parser_plugin_id=parser_plugin_id)
                     stored_file_tasks.append(task_id)
 
                     self.ap.logger.info(
@@ -189,21 +195,28 @@ class RuntimeKnowledgeBase(KnowledgeBaseInterface):
 
         return stored_file_tasks[0] if stored_file_tasks else ''
 
-    async def retrieve(self, query: str, top_k: int) -> list[rag_context.RetrievalResultEntry]:
-        embedding_model = await self.ap.model_mgr.get_embedding_model_by_uuid(
-            self.knowledge_base_entity.embedding_model_uuid
-        )
-        return await self.retriever.retrieve(self.knowledge_base_entity.uuid, query, embedding_model, top_k)
+    async def retrieve(self, query: str, settings: dict | None = None) -> list[rag_context.RetrievalResultEntry]:
+        # Merge stored retrieval_settings with per-request overrides
+        stored = self.knowledge_base_entity.retrieval_settings or {}
+        merged = {**stored, **(settings or {})}
+        if 'top_k' not in merged:
+            merged['top_k'] = 5  # fallback default
+
+        response = await self._retrieve(query, merged)
+
+        results_data = response.get('results', [])
+        entries = []
+        for r in results_data:
+            if isinstance(r, dict):
+                entries.append(rag_context.RetrievalResultEntry(**r))
+            elif isinstance(r, rag_context.RetrievalResultEntry):
+                entries.append(r)
+        return entries
 
     async def delete_file(self, file_id: str):
-        # delete vector
-        await self.ap.vector_db_mgr.vector_db.delete_by_file_id(self.knowledge_base_entity.uuid, file_id)
-
-        # delete chunk
-        await self.ap.persistence_mgr.execute_async(
-            sqlalchemy.delete(persistence_rag.Chunk).where(persistence_rag.Chunk.file_id == file_id)
-        )
+        await self._delete_document(file_id)
 
+        # Also cleanup DB record
         await self.ap.persistence_mgr.execute_async(
             sqlalchemy.delete(persistence_rag.File).where(persistence_rag.File.uuid == file_id)
         )
@@ -216,32 +229,295 @@ class RuntimeKnowledgeBase(KnowledgeBaseInterface):
         """Get the name of the knowledge base"""
         return self.knowledge_base_entity.name
 
-    def get_type(self) -> str:
-        """Get the type of knowledge base"""
-        return 'internal'
+    def get_knowledge_engine_plugin_id(self) -> str:
+        """Get the Knowledge Engine plugin ID"""
+        return self.knowledge_base_entity.knowledge_engine_plugin_id or ''
 
     async def dispose(self):
-        await self.ap.vector_db_mgr.vector_db.delete_collection(self.knowledge_base_entity.uuid)
+        """Dispose the knowledge base, notifying the plugin to cleanup."""
+        await self._on_kb_delete()
+
+    # ========== Plugin Communication Methods ==========
+
+    async def _on_kb_create(self) -> None:
+        """Notify plugin about KB creation."""
+        plugin_id = self.knowledge_base_entity.knowledge_engine_plugin_id
+        if not plugin_id:
+            return
+
+        try:
+            config = self.knowledge_base_entity.creation_settings or {}
+            self.ap.logger.info(
+                f'Calling RAG plugin {plugin_id}: on_knowledge_base_create(kb_id={self.knowledge_base_entity.uuid})'
+            )
+            await self.ap.plugin_connector.rag_on_kb_create(plugin_id, self.knowledge_base_entity.uuid, config)
+        except Exception as e:
+            self.ap.logger.error(f'Failed to notify plugin {plugin_id} on KB create: {e}')
+            raise
+
+    async def _on_kb_delete(self) -> None:
+        """Notify plugin about KB deletion."""
+        plugin_id = self.knowledge_base_entity.knowledge_engine_plugin_id
+        if not plugin_id:
+            return
+
+        try:
+            self.ap.logger.info(
+                f'Calling RAG plugin {plugin_id}: on_knowledge_base_delete(kb_id={self.knowledge_base_entity.uuid})'
+            )
+            await self.ap.plugin_connector.rag_on_kb_delete(plugin_id, self.knowledge_base_entity.uuid)
+        except Exception as e:
+            self.ap.logger.error(f'Failed to notify plugin {plugin_id} on KB delete: {e}')
+
+    async def _ingest_document(
+        self,
+        file_metadata: dict[str, Any],
+        storage_path: str,
+        parsed_content: dict[str, Any] | None = None,
+    ) -> dict[str, Any]:
+        """Call plugin to ingest document."""
+        kb = self.knowledge_base_entity
+        plugin_id = kb.knowledge_engine_plugin_id
+        if not plugin_id:
+            self.ap.logger.error(f'No RAG plugin ID configured for KB {kb.uuid}. Ingestion failed.')
+            raise ValueError('RAG Plugin ID required')
+
+        self.ap.logger.info(f'Calling RAG plugin {plugin_id}: ingest(doc={file_metadata.get("filename")})')
+
+        # Inject knowledge_base_id into file metadata as required by SDK schema
+        file_metadata['knowledge_base_id'] = kb.uuid
+
+        context_data = {
+            'file_object': {
+                'metadata': file_metadata,
+                'storage_path': storage_path,
+            },
+            'knowledge_base_id': kb.uuid,
+            'collection_id': kb.collection_id or kb.uuid,
+            'creation_settings': kb.creation_settings or {},
+            'parsed_content': parsed_content,
+        }
+
+        try:
+            result = await self.ap.plugin_connector.call_rag_ingest(plugin_id, context_data)
+            return result
+        except Exception as e:
+            self.ap.logger.error(f'Plugin ingestion failed: {e}')
+            raise
+
+    async def _retrieve(
+        self,
+        query: str,
+        settings: dict[str, Any],
+    ) -> dict[str, Any]:
+        """Call plugin to retrieve documents.
+
+        Raises:
+            ValueError: If no RAG plugin is configured for this KB.
+            Exception: If the plugin retrieval call fails.
+        """
+        kb = self.knowledge_base_entity
+        plugin_id = kb.knowledge_engine_plugin_id
+        if not plugin_id:
+            raise ValueError(f'No RAG plugin ID configured for KB {kb.uuid}. Retrieval failed.')
+
+        # Session context (e.g. session_name) stays in retrieval_settings
+        # for plugins that need it. Do NOT move them into filters, as filters
+        # are passed directly to vector_search by some plugins (e.g. LangRAG)
+        # and would cause empty results when the metadata field doesn't exist.
+        filters = settings.pop('filters', {})
+
+        retrieval_context = {
+            'query': query,
+            'knowledge_base_id': kb.uuid,
+            'collection_id': kb.collection_id or kb.uuid,
+            'retrieval_settings': settings,
+            'creation_settings': kb.creation_settings or {},
+            'filters': filters,
+        }
+
+        result = await self.ap.plugin_connector.call_rag_retrieve(
+            plugin_id,
+            retrieval_context,
+        )
+        return result
+
+    async def _delete_document(self, document_id: str) -> bool:
+        """Call plugin to delete document."""
+        kb = self.knowledge_base_entity
+        plugin_id = kb.knowledge_engine_plugin_id
+        if not plugin_id:
+            return False
+
+        self.ap.logger.info(f'Calling RAG plugin {plugin_id}: delete_document(doc_id={document_id})')
+
+        try:
+            return await self.ap.plugin_connector.call_rag_delete_document(plugin_id, document_id, kb.uuid)
+        except Exception as e:
+            self.ap.logger.error(f'Plugin document deletion failed: {e}')
+            return False
 
 
 class RAGManager:
     ap: app.Application
 
-    knowledge_bases: list[KnowledgeBaseInterface]
+    knowledge_bases: dict[str, KnowledgeBaseInterface]
 
     def __init__(self, ap: app.Application):
         self.ap = ap
-        self.knowledge_bases = []
+        self.knowledge_bases = {}
 
     async def initialize(self):
         await self.load_knowledge_bases_from_db()
 
+    async def get_all_knowledge_base_details(self) -> list[dict]:
+        """Get all knowledge bases with enriched Knowledge Engine details."""
+        # 1. Get raw KBs from DB
+        result = await self.ap.persistence_mgr.execute_async(sqlalchemy.select(persistence_rag.KnowledgeBase))
+        knowledge_bases = result.all()
+
+        # 2. Get all available Knowledge Engines for enrichment
+        engine_map = {}
+        if self.ap.plugin_connector.is_enable_plugin:
+            try:
+                engines = await self.ap.plugin_connector.list_knowledge_engines()
+                engine_map = {e['plugin_id']: e for e in engines}
+            except Exception as e:
+                self.ap.logger.warning(f'Failed to list Knowledge Engines: {e}')
+
+        # 3. Serialize and enrich
+        kb_list = []
+        for kb in knowledge_bases:
+            kb_dict = self.ap.persistence_mgr.serialize_model(persistence_rag.KnowledgeBase, kb)
+            self._enrich_kb_dict(kb_dict, engine_map)
+            kb_list.append(kb_dict)
+
+        return kb_list
+
+    async def get_knowledge_base_details(self, kb_uuid: str) -> dict | None:
+        """Get specific knowledge base with enriched Knowledge Engine details."""
+        result = await self.ap.persistence_mgr.execute_async(
+            sqlalchemy.select(persistence_rag.KnowledgeBase).where(persistence_rag.KnowledgeBase.uuid == kb_uuid)
+        )
+        kb = result.first()
+        if not kb:
+            return None
+
+        kb_dict = self.ap.persistence_mgr.serialize_model(persistence_rag.KnowledgeBase, kb)
+
+        # Fetch engines
+        engine_map = {}
+        if self.ap.plugin_connector.is_enable_plugin:
+            try:
+                engines = await self.ap.plugin_connector.list_knowledge_engines()
+                engine_map = {e['plugin_id']: e for e in engines}
+            except Exception as e:
+                self.ap.logger.warning(f'Failed to list Knowledge Engines: {e}')
+
+        self._enrich_kb_dict(kb_dict, engine_map)
+        return kb_dict
+
+    @staticmethod
+    def _to_i18n_name(name) -> dict:
+        """Ensure name is always an I18nObject-compatible dict.
+
+        If *name* is already a dict (with ``en_US`` / ``zh_Hans`` keys) it is
+        returned as-is.  A plain string is wrapped into an I18nObject so the
+        frontend ``extractI18nObject`` helper never receives an unexpected type.
+        """
+        if isinstance(name, dict):
+            return name
+        return {'en_US': str(name), 'zh_Hans': str(name)}
+
+    def _enrich_kb_dict(self, kb_dict: dict, engine_map: dict) -> None:
+        """Helper to inject engine info into KB dict."""
+        plugin_id = kb_dict.get('knowledge_engine_plugin_id')
+
+        # Default fallback structure — name must be I18nObject for frontend compatibility
+        fallback_name = self._to_i18n_name(plugin_id or 'Internal (Legacy)')
+        fallback_info = {
+            'plugin_id': plugin_id,
+            'name': fallback_name,
+            'capabilities': [],
+        }
+
+        if not plugin_id:
+            kb_dict['knowledge_engine'] = fallback_info
+            return
+
+        engine_info = engine_map.get(plugin_id)
+        if engine_info:
+            kb_dict['knowledge_engine'] = {
+                'plugin_id': plugin_id,
+                'name': self._to_i18n_name(engine_info.get('name', plugin_id)),
+                'capabilities': engine_info.get('capabilities', []),
+            }
+        else:
+            kb_dict['knowledge_engine'] = fallback_info
+
+    async def create_knowledge_base(
+        self,
+        name: str,
+        knowledge_engine_plugin_id: str,
+        creation_settings: dict,
+        retrieval_settings: dict | None = None,
+        description: str = '',
+    ) -> persistence_rag.KnowledgeBase:
+        """Create a new knowledge base using a RAG plugin."""
+        # Validate that the Knowledge Engine plugin exists
+        if self.ap.plugin_connector.is_enable_plugin:
+            try:
+                engines = await self.ap.plugin_connector.list_knowledge_engines()
+                engine_ids = [e.get('plugin_id') for e in engines]
+                if knowledge_engine_plugin_id not in engine_ids:
+                    raise ValueError(f'Knowledge Engine plugin {knowledge_engine_plugin_id} not found')
+            except ValueError:
+                raise
+            except Exception as e:
+                self.ap.logger.warning(f'Failed to validate Knowledge Engine plugin existence: {e}')
+
+        kb_uuid = str(uuid.uuid4())
+        # Use UUID as collection ID by default for isolation
+        collection_id = kb_uuid
+
+        kb_data = {
+            'uuid': kb_uuid,
+            'name': name,
+            'description': description,
+            'knowledge_engine_plugin_id': knowledge_engine_plugin_id,
+            'collection_id': collection_id,
+            'creation_settings': creation_settings,
+            'retrieval_settings': retrieval_settings or {},
+        }
+
+        # Create Entity
+        kb = persistence_rag.KnowledgeBase(**kb_data)
+
+        # Persist
+        await self.ap.persistence_mgr.execute_async(sqlalchemy.insert(persistence_rag.KnowledgeBase).values(kb_data))
+
+        # Load into Runtime
+        runtime_kb = await self.load_knowledge_base(kb)
+
+        # Notify Plugin — rollback DB record and runtime entry on failure
+        try:
+            await runtime_kb._on_kb_create()
+        except Exception:
+            self.knowledge_bases.pop(kb_uuid, None)
+            await self.ap.persistence_mgr.execute_async(
+                sqlalchemy.delete(persistence_rag.KnowledgeBase).where(persistence_rag.KnowledgeBase.uuid == kb_uuid)
+            )
+            raise
+
+        self.ap.logger.info(f'Created new Knowledge Base {name} ({kb_uuid}) using plugin {knowledge_engine_plugin_id}')
+        return kb
+
     async def load_knowledge_bases_from_db(self):
         self.ap.logger.info('Loading knowledge bases from db...')
 
-        self.knowledge_bases = []
+        self.knowledge_bases = {}
 
-        # Load internal knowledge bases
+        # Load knowledge bases
         result = await self.ap.persistence_mgr.execute_async(sqlalchemy.select(persistence_rag.KnowledgeBase))
         knowledge_bases = result.all()
 
@@ -253,86 +529,37 @@ class RAGManager:
                     f'Error loading knowledge base {knowledge_base.uuid}: {e}\n{traceback.format_exc()}'
                 )
 
-        # Load external knowledge bases
-        external_result = await self.ap.persistence_mgr.execute_async(
-            sqlalchemy.select(persistence_rag.ExternalKnowledgeBase)
-        )
-        external_kbs = external_result.all()
-
-        for external_kb in external_kbs:
-            try:
-                # Don't trigger sync during batch loading - will sync once after LangBot connects to runtime
-                await self.load_external_knowledge_base(external_kb, trigger_sync=False)
-            except Exception as e:
-                self.ap.logger.error(
-                    f'Error loading external knowledge base {external_kb.uuid}: {e}\n{traceback.format_exc()}'
-                )
-
     async def load_knowledge_base(
         self,
         knowledge_base_entity: persistence_rag.KnowledgeBase | sqlalchemy.Row | dict,
     ) -> RuntimeKnowledgeBase:
         if isinstance(knowledge_base_entity, sqlalchemy.Row):
+            # Safe access to _mapping for SQLAlchemy 1.4+
             knowledge_base_entity = persistence_rag.KnowledgeBase(**knowledge_base_entity._mapping)
         elif isinstance(knowledge_base_entity, dict):
-            knowledge_base_entity = persistence_rag.KnowledgeBase(**knowledge_base_entity)
+            # Filter out non-database fields (like knowledge_engine which is computed)
+            filtered_dict = {
+                k: v for k, v in knowledge_base_entity.items() if k in persistence_rag.KnowledgeBase.ALL_DB_FIELDS
+            }
+            knowledge_base_entity = persistence_rag.KnowledgeBase(**filtered_dict)
 
         runtime_knowledge_base = RuntimeKnowledgeBase(ap=self.ap, knowledge_base_entity=knowledge_base_entity)
 
         await runtime_knowledge_base.initialize()
 
-        self.knowledge_bases.append(runtime_knowledge_base)
+        self.knowledge_bases[runtime_knowledge_base.get_uuid()] = runtime_knowledge_base
 
         return runtime_knowledge_base
 
-    async def load_external_knowledge_base(
-        self,
-        external_kb_entity: persistence_rag.ExternalKnowledgeBase | sqlalchemy.Row | dict,
-        trigger_sync: bool = True,
-    ) -> ExternalKnowledgeBase:
-        """Load external knowledge base into runtime
-
-        Args:
-            external_kb_entity: External KB entity to load
-            trigger_sync: Whether to trigger sync after loading (default True for manual creation, False for batch loading)
-        """
-        if isinstance(external_kb_entity, sqlalchemy.Row):
-            external_kb_entity = persistence_rag.ExternalKnowledgeBase(**external_kb_entity._mapping)
-        elif isinstance(external_kb_entity, dict):
-            external_kb_entity = persistence_rag.ExternalKnowledgeBase(**external_kb_entity)
-
-        external_kb = ExternalKnowledgeBase(ap=self.ap, external_kb_entity=external_kb_entity)
-
-        await external_kb.initialize()
-
-        self.knowledge_bases.append(external_kb)
-
-        # Trigger sync to create the instance immediately (for manual creation)
-        # Skip sync during batch loading from DB to avoid multiple sync calls
-        if trigger_sync:
-            try:
-                await self.ap.plugin_connector.sync_polymorphic_component_instances()
-                self.ap.logger.info(f'Triggered sync after loading external KB {external_kb_entity.uuid}')
-            except Exception as e:
-                self.ap.logger.error(f'Failed to sync after loading external KB: {e}')
-
-        return external_kb
-
     async def get_knowledge_base_by_uuid(self, kb_uuid: str) -> KnowledgeBaseInterface | None:
-        for kb in self.knowledge_bases:
-            if kb.get_uuid() == kb_uuid:
-                return kb
-        return None
+        return self.knowledge_bases.get(kb_uuid)
 
     async def remove_knowledge_base_from_runtime(self, kb_uuid: str):
-        for kb in self.knowledge_bases:
-            if kb.get_uuid() == kb_uuid:
-                self.knowledge_bases.remove(kb)
-                return
+        self.knowledge_bases.pop(kb_uuid, None)
 
     async def delete_knowledge_base(self, kb_uuid: str):
-        for kb in self.knowledge_bases:
-            if kb.get_uuid() == kb_uuid:
-                await kb.dispose()
-                self.knowledge_bases.remove(kb)
-                return
+        kb = self.knowledge_bases.pop(kb_uuid, None)
+        if kb is not None:
+            await kb.dispose()
+        else:
+            self.ap.logger.warning(f'Knowledge base {kb_uuid} not found in runtime, skipping plugin notification')

src/langbot/pkg/rag/knowledge/services/base_service.py

@@ -1,15 +0,0 @@
-# 封装异步操作
-import asyncio
-
-
-class BaseService:
-    def __init__(self):
-        pass
-
-    async def _run_sync(self, func, *args, **kwargs):
-        """
-        在单独的线程中运行同步函数。
-        如果第一个参数是 session，则在 to_thread 中获取新的 session。
-        """
-
-        return await asyncio.to_thread(func, *args, **kwargs)

src/langbot/pkg/rag/knowledge/services/chunker.py

@@ -1,49 +0,0 @@
-from __future__ import annotations
-
-import json
-from typing import List
-from langbot.pkg.rag.knowledge.services import base_service
-from langbot.pkg.core import app
-from langchain_text_splitters import RecursiveCharacterTextSplitter
-
-
-class Chunker(base_service.BaseService):
-    """
-    A class for splitting long texts into smaller, overlapping chunks.
-    """
-
-    def __init__(self, ap: app.Application, chunk_size: int = 500, chunk_overlap: int = 50):
-        self.ap = ap
-        self.chunk_size = chunk_size
-        self.chunk_overlap = chunk_overlap
-        if self.chunk_overlap >= self.chunk_size:
-            self.ap.logger.warning(
-                'Chunk overlap is greater than or equal to chunk size. This may lead to empty or malformed chunks.'
-            )
-
-    def _split_text_sync(self, text: str) -> List[str]:
-        """
-        Synchronously splits a long text into chunks with specified overlap.
-        This is a CPU-bound operation, intended to be run in a separate thread.
-        """
-        if not text:
-            return []
-
-        text_splitter = RecursiveCharacterTextSplitter(
-            chunk_size=self.chunk_size,
-            chunk_overlap=self.chunk_overlap,
-            length_function=len,
-            is_separator_regex=False,
-        )
-        return text_splitter.split_text(text)
-
-    async def chunk(self, text: str) -> List[str]:
-        """
-        Asynchronously chunks a given text into smaller pieces.
-        """
-        self.ap.logger.info(f'Chunking text (length: {len(text)})...')
-        # Run the synchronous splitting logic in a separate thread
-        chunks = await self._run_sync(self._split_text_sync, text)
-        self.ap.logger.info(f'Text chunked into {len(chunks)} pieces.')
-        self.ap.logger.debug(f'Chunks: {json.dumps(chunks, indent=4, ensure_ascii=False)}')
-        return chunks

src/langbot/pkg/rag/knowledge/services/embedder.py

@@ -1,55 +0,0 @@
-from __future__ import annotations
-import uuid
-from typing import List
-from langbot.pkg.rag.knowledge.services.base_service import BaseService
-from langbot.pkg.entity.persistence import rag as persistence_rag
-from langbot.pkg.core import app
-from langbot.pkg.provider.modelmgr.requester import RuntimeEmbeddingModel
-import sqlalchemy
-
-
-class Embedder(BaseService):
-    def __init__(self, ap: app.Application) -> None:
-        super().__init__()
-        self.ap = ap
-
-    async def embed_and_store(
-        self, kb_id: str, file_id: str, chunks: List[str], embedding_model: RuntimeEmbeddingModel
-    ) -> list[persistence_rag.Chunk]:
-        # save chunk to db
-        chunk_entities: list[persistence_rag.Chunk] = []
-        chunk_ids: list[str] = []
-
-        for chunk_text in chunks:
-            chunk_uuid = str(uuid.uuid4())
-            chunk_ids.append(chunk_uuid)
-            chunk_entity = persistence_rag.Chunk(uuid=chunk_uuid, file_id=file_id, text=chunk_text)
-            chunk_entities.append(chunk_entity)
-
-        chunk_dicts = [
-            self.ap.persistence_mgr.serialize_model(persistence_rag.Chunk, chunk) for chunk in chunk_entities
-        ]
-
-        await self.ap.persistence_mgr.execute_async(sqlalchemy.insert(persistence_rag.Chunk).values(chunk_dicts))
-
-        # get embeddings (batch size limit: 64 for OpenAI)
-        MAX_BATCH_SIZE = 64
-        embeddings_list: list[list[float]] = []
-
-        for i in range(0, len(chunks), MAX_BATCH_SIZE):
-            batch = chunks[i : i + MAX_BATCH_SIZE]
-            batch_embeddings = await embedding_model.provider.invoke_embedding(
-                model=embedding_model,
-                input_text=batch,
-                extra_args={},  # TODO: add extra args
-                knowledge_base_id=kb_id,
-                call_type='embedding',
-            )
-            embeddings_list.extend(batch_embeddings)
-
-        # save embeddings to vdb
-        await self.ap.vector_db_mgr.vector_db.add_embeddings(kb_id, chunk_ids, embeddings_list, chunk_dicts)
-
-        self.ap.logger.info(f'Successfully saved {len(chunk_entities)} embeddings to Knowledge Base.')
-
-        return chunk_entities

src/langbot/pkg/rag/knowledge/services/parser.py

@@ -1,291 +0,0 @@
-from __future__ import annotations
-
-import PyPDF2
-import io
-from docx import Document
-import chardet
-from typing import Union, Callable, Any
-import markdown
-from bs4 import BeautifulSoup
-import re
-import asyncio  # Import asyncio for async operations
-from langbot.pkg.core import app
-
-
-class FileParser:
-    """
-    A robust file parser class to extract text content from various document formats.
-    It supports TXT, PDF, DOCX, XLSX, CSV, Markdown, HTML, and EPUB files.
-    All core file reading operations are designed to be run synchronously in a thread pool
-    to avoid blocking the asyncio event loop.
-    """
-
-    def __init__(self, ap: app.Application):
-        self.ap = ap
-
-    async def _run_sync(self, sync_func: Callable, *args: Any, **kwargs: Any) -> Any:
-        """
-        Runs a synchronous function in a separate thread to prevent blocking the event loop.
-        This is a general utility method for wrapping blocking I/O operations.
-        """
-        try:
-            return await asyncio.to_thread(sync_func, *args, **kwargs)
-        except Exception as e:
-            self.ap.logger.error(f'Error running synchronous function {sync_func.__name__}: {e}')
-            raise
-
-    async def parse(self, file_name: str, extension: str) -> Union[str, None]:
-        """
-        Parses the file based on its extension and returns the extracted text content.
-        This is the main asynchronous entry point for parsing.
-
-        Args:
-            file_name (str): The name of the file to be parsed, get from ap.storage_mgr
-
-        Returns:
-            Union[str, None]: The extracted text content as a single string, or None if parsing fails.
-        """
-
-        file_extension = extension.lower()
-        parser_method = getattr(self, f'_parse_{file_extension}', None)
-
-        if parser_method is None:
-            self.ap.logger.error(f'Unsupported file format: {file_extension} for file {file_name}')
-            return None
-
-        try:
-            # Pass file_path to the specific parser methods
-            return await parser_method(file_name)
-        except Exception as e:
-            self.ap.logger.error(f'Failed to parse {file_extension} file {file_name}: {e}')
-            return None
-
-    # --- Helper for reading files with encoding detection ---
-    async def _read_file_content(self, file_name: str) -> Union[str, bytes]:
-        """
-        Reads a file with automatic encoding detection, ensuring the synchronous
-        file read operation runs in a separate thread.
-        """
-
-        # def _read_sync():
-        #     with open(file_path, 'rb') as file:
-        #         raw_data = file.read()
-        #         detected = chardet.detect(raw_data)
-        #         encoding = detected['encoding'] or 'utf-8'
-
-        #     if mode == 'r':
-        #         return raw_data.decode(encoding, errors='ignore')
-        #     return raw_data  # For binary mode
-
-        # return await self._run_sync(_read_sync)
-        file_bytes = await self.ap.storage_mgr.storage_provider.load(file_name)
-
-        detected = chardet.detect(file_bytes)
-        encoding = detected['encoding'] or 'utf-8'
-
-        return file_bytes.decode(encoding, errors='ignore')
-
-    # --- Specific Parser Methods ---
-
-    async def _parse_txt(self, file_name: str) -> str:
-        """Parses a TXT file and returns its content."""
-        self.ap.logger.info(f'Parsing TXT file: {file_name}')
-        return await self._read_file_content(file_name)
-
-    async def _parse_pdf(self, file_name: str) -> str:
-        """Parses a PDF file and returns its text content."""
-        self.ap.logger.info(f'Parsing PDF file: {file_name}')
-
-        # def _parse_pdf_sync():
-        #     text_content = []
-        #     with open(file_name, 'rb') as file:
-        #         pdf_reader = PyPDF2.PdfReader(file)
-        #         for page in pdf_reader.pages:
-        #             text = page.extract_text()
-        #             if text:
-        #                 text_content.append(text)
-        #     return '\n'.join(text_content)
-
-        # return await self._run_sync(_parse_pdf_sync)
-
-        pdf_bytes = await self.ap.storage_mgr.storage_provider.load(file_name)
-
-        def _parse_pdf_sync():
-            pdf_reader = PyPDF2.PdfReader(io.BytesIO(pdf_bytes))
-            text_content = []
-            for page in pdf_reader.pages:
-                text = page.extract_text()
-                if text:
-                    text_content.append(text)
-            return '\n'.join(text_content)
-
-        return await self._run_sync(_parse_pdf_sync)
-
-    async def _parse_docx(self, file_name: str) -> str:
-        """Parses a DOCX file and returns its text content."""
-        self.ap.logger.info(f'Parsing DOCX file: {file_name}')
-
-        docx_bytes = await self.ap.storage_mgr.storage_provider.load(file_name)
-
-        def _parse_docx_sync():
-            doc = Document(io.BytesIO(docx_bytes))
-            text_content = [paragraph.text for paragraph in doc.paragraphs if paragraph.text.strip()]
-            return '\n'.join(text_content)
-
-        return await self._run_sync(_parse_docx_sync)
-
-    async def _parse_doc(self, file_name: str) -> str:
-        """Handles .doc files, explicitly stating lack of direct support."""
-        self.ap.logger.warning(f'Direct .doc parsing is not supported for {file_name}. Please convert to .docx first.')
-        raise NotImplementedError('Direct .doc parsing not supported. Please convert to .docx first.')
-
-    # async def _parse_xlsx(self, file_name: str) -> str:
-    #     """Parses an XLSX file, returning text from all sheets."""
-    #     self.ap.logger.info(f'Parsing XLSX file: {file_name}')
-
-    #     xlsx_bytes = await self.ap.storage_mgr.storage_provider.load(file_name)
-
-    #     def _parse_xlsx_sync():
-    #         excel_file = pd.ExcelFile(io.BytesIO(xlsx_bytes))
-    #         all_sheet_content = []
-    #         for sheet_name in excel_file.sheet_names:
-    #             df = pd.read_excel(io.BytesIO(xlsx_bytes), sheet_name=sheet_name)
-    #             sheet_text = f'--- Sheet: {sheet_name} ---\n{df.to_string(index=False)}\n'
-    #             all_sheet_content.append(sheet_text)
-    #         return '\n'.join(all_sheet_content)
-
-    #     return await self._run_sync(_parse_xlsx_sync)
-
-    # async def _parse_csv(self, file_name: str) -> str:
-    #     """Parses a CSV file and returns its content as a string."""
-    #     self.ap.logger.info(f'Parsing CSV file: {file_name}')
-
-    #     csv_bytes = await self.ap.storage_mgr.storage_provider.load(file_name)
-
-    #     def _parse_csv_sync():
-    #         # pd.read_csv can often detect encoding, but explicit detection is safer
-    #         # raw_data = self._read_file_content(
-    #         #     file_name, mode='rb'
-    #         # )  # Note: this will need to be await outside this sync function
-    #         # _ = raw_data
-    #         # For simplicity, we'll let pandas handle encoding internally after a raw read.
-    #         # A more robust solution might pass encoding directly to pd.read_csv after detection.
-    #         detected = chardet.detect(io.BytesIO(csv_bytes))
-    #         encoding = detected['encoding'] or 'utf-8'
-    #         df = pd.read_csv(io.BytesIO(csv_bytes), encoding=encoding)
-    #         return df.to_string(index=False)
-
-    #     return await self._run_sync(_parse_csv_sync)
-
-    async def _parse_md(self, file_name: str) -> str:
-        """Parses a Markdown file, converting it to structured plain text."""
-        self.ap.logger.info(f'Parsing Markdown file: {file_name}')
-
-        md_bytes = await self.ap.storage_mgr.storage_provider.load(file_name)
-
-        def _parse_markdown_sync():
-            md_content = io.BytesIO(md_bytes).read().decode('utf-8', errors='ignore')
-            html_content = markdown.markdown(
-                md_content, extensions=['extra', 'codehilite', 'tables', 'toc', 'fenced_code']
-            )
-            soup = BeautifulSoup(html_content, 'html.parser')
-            text_parts = []
-            for element in soup.children:
-                if element.name in ['h1', 'h2', 'h3', 'h4', 'h5', 'h6']:
-                    level = int(element.name[1])
-                    text_parts.append('#' * level + ' ' + element.get_text().strip())
-                elif element.name == 'p':
-                    text = element.get_text().strip()
-                    if text:
-                        text_parts.append(text)
-                elif element.name in ['ul', 'ol']:
-                    for li in element.find_all('li'):
-                        text_parts.append(f'* {li.get_text().strip()}')
-                elif element.name == 'pre':
-                    code_block = element.get_text().strip()
-                    if code_block:
-                        text_parts.append(f'```\n{code_block}\n```')
-                elif element.name == 'table':
-                    table_str = self._extract_table_to_markdown_sync(element)  # Call sync helper
-                    if table_str:
-                        text_parts.append(table_str)
-                elif element.name:
-                    text = element.get_text(separator=' ', strip=True)
-                    if text:
-                        text_parts.append(text)
-            cleaned_text = re.sub(r'\n\s*\n', '\n\n', '\n'.join(text_parts))
-            return cleaned_text.strip()
-
-        return await self._run_sync(_parse_markdown_sync)
-
-    async def _parse_html(self, file_name: str) -> str:
-        """Parses an HTML file, extracting structured plain text."""
-        self.ap.logger.info(f'Parsing HTML file: {file_name}')
-
-        html_bytes = await self.ap.storage_mgr.storage_provider.load(file_name)
-
-        def _parse_html_sync():
-            html_content = io.BytesIO(html_bytes).read().decode('utf-8', errors='ignore')
-            soup = BeautifulSoup(html_content, 'html.parser')
-            for script_or_style in soup(['script', 'style']):
-                script_or_style.decompose()
-            text_parts = []
-            for element in soup.body.children if soup.body else soup.children:
-                if element.name in ['h1', 'h2', 'h3', 'h4', 'h5', 'h6']:
-                    level = int(element.name[1])
-                    text_parts.append('#' * level + ' ' + element.get_text().strip())
-                elif element.name == 'p':
-                    text = element.get_text().strip()
-                    if text:
-                        text_parts.append(text)
-                elif element.name in ['ul', 'ol']:
-                    for li in element.find_all('li'):
-                        text = li.get_text().strip()
-                        if text:
-                            text_parts.append(f'* {text}')
-                elif element.name == 'table':
-                    table_str = self._extract_table_to_markdown_sync(element)  # Call sync helper
-                    if table_str:
-                        text_parts.append(table_str)
-                elif element.name:
-                    text = element.get_text(separator=' ', strip=True)
-                    if text:
-                        text_parts.append(text)
-            cleaned_text = re.sub(r'\n\s*\n', '\n\n', '\n'.join(text_parts))
-            return cleaned_text.strip()
-
-        return await self._run_sync(_parse_html_sync)
-
-    def _add_toc_items_sync(self, toc_list: list, text_content: list, level: int):
-        """Recursively adds TOC items to text_content (synchronous helper)."""
-        indent = '  ' * level
-        for item in toc_list:
-            if isinstance(item, tuple):
-                chapter, subchapters = item
-                text_content.append(f'{indent}- {chapter.title}')
-                self._add_toc_items_sync(subchapters, text_content, level + 1)
-            else:
-                text_content.append(f'{indent}- {item.title}')
-
-    def _extract_table_to_markdown_sync(self, table_element: BeautifulSoup) -> str:
-        """Helper to convert a BeautifulSoup table element into a Markdown table string (synchronous)."""
-        headers = [th.get_text().strip() for th in table_element.find_all('th')]
-        rows = []
-        for tr in table_element.find_all('tr'):
-            cells = [td.get_text().strip() for td in tr.find_all('td')]
-            if cells:
-                rows.append(cells)
-
-        if not headers and not rows:
-            return ''
-
-        table_lines = []
-        if headers:
-            table_lines.append(' | '.join(headers))
-            table_lines.append(' | '.join(['---'] * len(headers)))
-
-        for row_cells in rows:
-            padded_cells = row_cells + [''] * (len(headers) - len(row_cells)) if headers else row_cells
-            table_lines.append(' | '.join(padded_cells))
-
-        return '\n'.join(table_lines)

src/langbot/pkg/rag/knowledge/services/retriever.py

@@ -1,53 +0,0 @@
-from __future__ import annotations
-
-from . import base_service
-from ....core import app
-from ....provider.modelmgr.requester import RuntimeEmbeddingModel
-from langbot_plugin.api.entities.builtin.rag import context as rag_context
-from langbot_plugin.api.entities.builtin.provider.message import ContentElement
-
-
-class Retriever(base_service.BaseService):
-    def __init__(self, ap: app.Application):
-        super().__init__()
-        self.ap = ap
-
-    async def retrieve(
-        self, kb_id: str, query: str, embedding_model: RuntimeEmbeddingModel, k: int = 5
-    ) -> list[rag_context.RetrievalResultEntry]:
-        self.ap.logger.info(
-            f"Retrieving for query: '{query[:10]}' with k={k} using {embedding_model.model_entity.uuid}"
-        )
-
-        query_embedding: list[float] = await embedding_model.provider.invoke_embedding(
-            model=embedding_model,
-            input_text=[query],
-            extra_args={},  # TODO: add extra args
-            knowledge_base_id=kb_id,
-            query_text=query,
-            call_type='retrieve',
-        )
-
-        vector_results = await self.ap.vector_db_mgr.vector_db.search(kb_id, query_embedding[0], k)
-
-        # 'ids' shape mirrors the Chroma-style response contract for compatibility
-        matched_vector_ids = vector_results.get('ids', [[]])[0]
-        distances = vector_results.get('distances', [[]])[0]
-        vector_metadatas = vector_results.get('metadatas', [[]])[0]
-
-        if not matched_vector_ids:
-            self.ap.logger.info('No relevant chunks found in vector database.')
-            return []
-
-        result: list[rag_context.RetrievalResultEntry] = []
-
-        for i, id in enumerate(matched_vector_ids):
-            entry = rag_context.RetrievalResultEntry(
-                id=id,
-                content=[ContentElement.from_text(vector_metadatas[i].get('text', ''))],
-                metadata=vector_metadatas[i],
-                distance=distances[i],
-            )
-            result.append(entry)
-
-        return result

src/langbot/pkg/rag/service/__init__.py

@@ -0,0 +1 @@
+from .runtime import RAGRuntimeService as RAGRuntimeService

src/langbot/pkg/rag/service/runtime.py

@@ -0,0 +1,89 @@
+from __future__ import annotations
+
+import posixpath
+from typing import Any
+from langbot.pkg.core import app
+
+
+class RAGRuntimeService:
+    """Service to handle RAG-related requests from plugins (Runtime).
+
+    This service acts as the bridge between plugin RPC requests and
+    LangBot's infrastructure (embedding models, vector databases, file storage).
+    """
+
+    def __init__(self, ap: app.Application):
+        self.ap = ap
+
+    async def vector_upsert(
+        self,
+        collection_id: str,
+        vectors: list[list[float]],
+        ids: list[str],
+        metadata: list[dict[str, Any]] | None = None,
+        documents: list[str] | None = None,
+    ) -> None:
+        """Handle VECTOR_UPSERT action."""
+        metadatas = metadata if metadata else [{} for _ in vectors]
+        await self.ap.vector_db_mgr.upsert(
+            collection_name=collection_id,
+            vectors=vectors,
+            ids=ids,
+            metadata=metadatas,
+            documents=documents,
+        )
+
+    async def vector_search(
+        self,
+        collection_id: str,
+        query_vector: list[float],
+        top_k: int,
+        filters: dict[str, Any] | None = None,
+        search_type: str = 'vector',
+        query_text: str = '',
+    ) -> list[dict[str, Any]]:
+        """Handle VECTOR_SEARCH action."""
+        return await self.ap.vector_db_mgr.search(
+            collection_name=collection_id,
+            query_vector=query_vector,
+            limit=top_k,
+            filter=filters,
+            search_type=search_type,
+            query_text=query_text,
+        )
+
+    async def vector_delete(
+        self, collection_id: str, file_ids: list[str] | None = None, filters: dict[str, Any] | None = None
+    ) -> int:
+        """Handle VECTOR_DELETE action.
+
+        Deletes vectors associated with the given file IDs from the collection.
+        Each file_id corresponds to a document whose vectors will be removed.
+
+        Args:
+            collection_id: The collection to delete from.
+            file_ids: File IDs whose associated vectors should be deleted.
+                Each file_id maps to a set of vectors stored with that file_id
+                in their metadata.
+            filters: Filter-based deletion (not yet supported, will raise).
+        """
+        count = 0
+        if file_ids:
+            await self.ap.vector_db_mgr.delete_by_file_id(collection_name=collection_id, file_ids=file_ids)
+            count = len(file_ids)
+        elif filters:
+            count = await self.ap.vector_db_mgr.delete_by_filter(collection_name=collection_id, filter=filters)
+        return count
+
+    async def get_file_stream(self, storage_path: str) -> bytes:
+        """Handle GET_KNOWLEDEGE_FILE_STREAM action.
+
+        Uses the storage manager abstraction to load file content,
+        regardless of the underlying storage provider.
+        """
+        # Validate storage_path to prevent path traversal
+        normalized = posixpath.normpath(storage_path)
+        if normalized.startswith('/') or '..' in normalized.split('/'):
+            raise ValueError('Invalid storage path')
+        content_bytes = await self.ap.storage_mgr.storage_provider.load(normalized)
+        return content_bytes if content_bytes else b''

src/langbot/pkg/storage/mgr.py

@@ -3,7 +3,7 @@ from __future__ import annotations
 
 from ..core import app
 from . import provider
-from .providers import localstorage, s3storage
+from .providers import localstorage
 
 
 class StorageMgr:
@@ -21,6 +21,8 @@ class StorageMgr:
         storage_type = storage_config.get('use', 'local')
 
         if storage_type == 's3':
+            from .providers import s3storage
+
             self.storage_provider = s3storage.S3StorageProvider(self.ap)
             self.ap.logger.info('Initialized S3 storage backend.')
         else:

src/langbot/pkg/storage/provider.py

@@ -43,6 +43,13 @@ class StorageProvider(abc.ABC):
     ):
         pass
 
+    @abc.abstractmethod
+    async def size(
+        self,
+        key: str,
+    ) -> int:
+        pass
+
     @abc.abstractmethod
     async def delete_dir_recursive(
         self,

src/langbot/pkg/storage/providers/localstorage.py

@@ -47,6 +47,12 @@ class LocalStorageProvider(provider.StorageProvider):
     ):
         os.remove(os.path.join(LOCAL_STORAGE_PATH, f'{key}'))
 
+    async def size(
+        self,
+        key: str,
+    ) -> int:
+        return os.path.getsize(os.path.join(LOCAL_STORAGE_PATH, f'{key}'))
+
     async def delete_dir_recursive(
         self,
         dir_path: str,

src/langbot/pkg/storage/providers/s3storage.py

@@ -117,6 +117,21 @@ class S3StorageProvider(provider.StorageProvider):
             self.ap.logger.error(f'Failed to delete from S3: {e}')
             raise
 
+    async def size(
+        self,
+        key: str,
+    ) -> int:
+        """Get object size from S3 without downloading it"""
+        try:
+            response = self.s3_client.head_object(
+                Bucket=self.bucket_name,
+                Key=key,
+            )
+            return response['ContentLength']
+        except Exception as e:
+            self.ap.logger.error(f'Failed to get size from S3: {e}')
+            raise
+
     async def delete_dir_recursive(
         self,
         dir_path: str,

src/langbot/pkg/telemetry/telemetry.py

@@ -60,7 +60,7 @@ class TelemetryManager:
                     except Exception:
                         sanitized['query_id'] = str(sanitized.get('query_id', ''))
 
-                for sfield in ('adapter', 'runner', 'model_name', 'version', 'error', 'timestamp'):
+                for sfield in ('adapter', 'runner', 'runner_category', 'model_name', 'version', 'error', 'timestamp'):
                     v = sanitized.get(sfield)
                     sanitized[sfield] = '' if v is None else str(v)
 

src/langbot/pkg/utils/constants.py

@@ -2,7 +2,7 @@ import langbot
 
 semantic_version = f'v{langbot.__version__}'
 
-required_database_version = 19
+required_database_version = 24
 """Tag the version of the database schema, used to check if the database needs to be migrated"""
 
 debug_mode = False

src/langbot/pkg/utils/httpclient.py

@@ -0,0 +1,43 @@
+"""Shared aiohttp.ClientSession to avoid repeated SSL context creation.
+
+Each call to `aiohttp.ClientSession()` creates a new `TCPConnector` which in turn
+creates a new `ssl.SSLContext` and loads all system root certificates. This is
+extremely expensive in both CPU and memory (~270MB total allocations observed via
+memray profiling).
+
+This module provides a shared session pool so that all HTTP client code in LangBot
+reuses the same underlying SSL context and connection pool.
+"""
+
+from __future__ import annotations
+
+import aiohttp
+
+_sessions: dict[str, aiohttp.ClientSession] = {}
+
+
+def get_session(*, trust_env: bool = False) -> aiohttp.ClientSession:
+    """Get or create a shared aiohttp.ClientSession.
+
+    Args:
+        trust_env: Whether to trust environment variables for proxy settings.
+
+    Returns:
+        A shared aiohttp.ClientSession instance.
+    """
+    key = f'trust_env={trust_env}'
+
+    session = _sessions.get(key)
+    if session is None or session.closed:
+        session = aiohttp.ClientSession(trust_env=trust_env)
+        _sessions[key] = session
+
+    return session
+
+
+async def close_all():
+    """Close all shared sessions. Call on application shutdown."""
+    for session in _sessions.values():
+        if not session.closed:
+            await session.close()
+    _sessions.clear()

src/langbot/pkg/utils/image.py

@@ -5,6 +5,8 @@ from urllib.parse import urlparse, parse_qs
 import ssl
 
 import aiohttp
+
+from langbot.pkg.utils import httpclient
 import PIL.Image
 import httpx
 
@@ -47,53 +49,54 @@ async def get_gewechat_image_base64(
     )
 
     try:
-        async with aiohttp.ClientSession(timeout=timeout) as session:
-            # 获取图片下载链接
-            try:
-                async with session.post(
-                    f'{gewechat_url}/v2/api/message/downloadImage',
-                    headers=headers,
-                    json={'appId': app_id, 'type': image_type, 'xml': xml_content},
-                ) as response:
-                    if response.status != 200:
-                        # print(response)
-                        raise Exception(f'获取gewechat图片下载失败: {await response.text()}')
+        session = httpclient.get_session()
+        # 获取图片下载链接
+        try:
+            async with session.post(
+                f'{gewechat_url}/v2/api/message/downloadImage',
+                headers=headers,
+                json={'appId': app_id, 'type': image_type, 'xml': xml_content},
+                timeout=timeout,
+            ) as response:
+                if response.status != 200:
+                    # print(response)
+                    raise Exception(f'获取gewechat图片下载失败: {await response.text()}')
 
-                    resp_data = await response.json()
-                    if resp_data.get('ret') != 200:
-                        raise Exception(f'获取gewechat图片下载链接失败: {resp_data}')
+                resp_data = await response.json()
+                if resp_data.get('ret') != 200:
+                    raise Exception(f'获取gewechat图片下载链接失败: {resp_data}')
 
-                    file_url = resp_data['data']['fileUrl']
-            except asyncio.TimeoutError:
-                raise Exception('获取图片下载链接超时')
-            except aiohttp.ClientError as e:
-                raise Exception(f'获取图片下载链接网络错误: {str(e)}')
+                file_url = resp_data['data']['fileUrl']
+        except asyncio.TimeoutError:
+            raise Exception('获取图片下载链接超时')
+        except aiohttp.ClientError as e:
+            raise Exception(f'获取图片下载链接网络错误: {str(e)}')
 
-            # 解析原始URL并替换端口
-            base_url = gewechat_file_url
-            download_url = f'{base_url}/download/{file_url}'
+        # 解析原始URL并替换端口
+        base_url = gewechat_file_url
+        download_url = f'{base_url}/download/{file_url}'
 
-            # 下载图片
-            try:
-                async with session.get(download_url) as img_response:
-                    if img_response.status != 200:
-                        raise Exception(f'下载图片失败: {await img_response.text()}, URL: {download_url}')
+        # 下载图片
+        try:
+            async with session.get(download_url) as img_response:
+                if img_response.status != 200:
+                    raise Exception(f'下载图片失败: {await img_response.text()}, URL: {download_url}')
 
-                    image_data = await img_response.read()
+                image_data = await img_response.read()
 
-                    content_type = img_response.headers.get('Content-Type', '')
-                    if content_type:
-                        image_format = content_type.split('/')[-1]
-                    else:
-                        image_format = file_url.split('.')[-1]
+                content_type = img_response.headers.get('Content-Type', '')
+                if content_type:
+                    image_format = content_type.split('/')[-1]
+                else:
+                    image_format = file_url.split('.')[-1]
 
-                    base64_str = base64.b64encode(image_data).decode('utf-8')
+                base64_str = base64.b64encode(image_data).decode('utf-8')
 
-                    return base64_str, image_format
-            except asyncio.TimeoutError:
-                raise Exception(f'下载图片超时, URL: {download_url}')
-            except aiohttp.ClientError as e:
-                raise Exception(f'下载图片网络错误: {str(e)}, URL: {download_url}')
+                return base64_str, image_format
+        except asyncio.TimeoutError:
+            raise Exception(f'下载图片超时, URL: {download_url}')
+        except aiohttp.ClientError as e:
+            raise Exception(f'下载图片网络错误: {str(e)}, URL: {download_url}')
     except Exception as e:
         raise Exception(f'获取图片失败: {str(e)}') from e
 
@@ -104,24 +107,24 @@ async def get_wecom_image_base64(pic_url: str) -> tuple[str, str]:
     :param pic_url: 企业微信图片URL
     :return: (base64_str, image_format)
     """
-    async with aiohttp.ClientSession() as session:
-        async with session.get(pic_url) as response:
-            if response.status != 200:
-                raise Exception(f'Failed to download image: {response.status}')
+    session = httpclient.get_session()
+    async with session.get(pic_url) as response:
+        if response.status != 200:
+            raise Exception(f'Failed to download image: {response.status}')
 
-            # 读取图片数据
-            image_data = await response.read()
+        # 读取图片数据
+        image_data = await response.read()
 
-            # 获取图片格式
-            content_type = response.headers.get('Content-Type', '')
-            image_format = content_type.split('/')[-1]  # 例如 'image/jpeg' -> 'jpeg'
+        # 获取图片格式
+        content_type = response.headers.get('Content-Type', '')
+        image_format = content_type.split('/')[-1]  # 例如 'image/jpeg' -> 'jpeg'
 
-            # 转换为 base64
-            import base64
+        # 转换为 base64
+        import base64
 
-            image_base64 = base64.b64encode(image_data).decode('utf-8')
+        image_base64 = base64.b64encode(image_data).decode('utf-8')
 
-            return image_base64, image_format
+        return image_base64, image_format
 
 
 async def get_qq_official_image_base64(pic_url: str, content_type: str) -> tuple[str, str]:
@@ -152,21 +155,19 @@ async def get_qq_image_bytes(image_url: str, query: dict = {}) -> tuple[bytes, s
     ssl_context = ssl.create_default_context()
     ssl_context.check_hostname = False
     ssl_context.verify_mode = ssl.CERT_NONE
-    async with aiohttp.ClientSession(trust_env=False) as session:
-        async with session.get(
-            image_url, params=query, ssl=ssl_context, timeout=aiohttp.ClientTimeout(total=30.0)
-        ) as resp:
-            resp.raise_for_status()
-            file_bytes = await resp.read()
-            content_type = resp.headers.get('Content-Type')
-            if not content_type:
-                image_format = 'jpeg'
-            elif not content_type.startswith('image/'):
-                pil_img = PIL.Image.open(io.BytesIO(file_bytes))
-                image_format = pil_img.format.lower()
-            else:
-                image_format = content_type.split('/')[-1]
-            return file_bytes, image_format
+    session = httpclient.get_session()
+    async with session.get(image_url, params=query, ssl=ssl_context, timeout=aiohttp.ClientTimeout(total=30.0)) as resp:
+        resp.raise_for_status()
+        file_bytes = await resp.read()
+        content_type = resp.headers.get('Content-Type')
+        if not content_type:
+            image_format = 'jpeg'
+        elif not content_type.startswith('image/'):
+            pil_img = PIL.Image.open(io.BytesIO(file_bytes))
+            image_format = pil_img.format.lower()
+        else:
+            image_format = content_type.split('/')[-1]
+        return file_bytes, image_format
 
 
 async def qq_image_url_to_base64(image_url: str) -> typing.Tuple[str, str]:
@@ -204,11 +205,11 @@ async def extract_b64_and_format(image_base64_data: str) -> typing.Tuple[str, st
 async def get_slack_image_to_base64(pic_url: str, bot_token: str):
     headers = {'Authorization': f'Bearer {bot_token}'}
     try:
-        async with aiohttp.ClientSession() as session:
-            async with session.get(pic_url, headers=headers) as resp:
-                mime_type = resp.headers.get('Content-Type', 'application/octet-stream')
-                file_bytes = await resp.read()
-                base64_str = base64.b64encode(file_bytes).decode('utf-8')
-            return f'data:{mime_type};base64,{base64_str}'
+        session = httpclient.get_session()
+        async with session.get(pic_url, headers=headers) as resp:
+            mime_type = resp.headers.get('Content-Type', 'application/octet-stream')
+            file_bytes = await resp.read()
+            base64_str = base64.b64encode(file_bytes).decode('utf-8')
+        return f'data:{mime_type};base64,{base64_str}'
     except Exception as e:
         raise (e)

src/langbot/pkg/utils/runner.py

@@ -0,0 +1,105 @@
+from __future__ import annotations
+
+from urllib.parse import urlparse
+
+
+class RunnerCategory:
+    LOCAL = 'local'
+    CLOUD = 'cloud'
+    UNKNOWN = 'unknown'
+
+
+CLOUD_DOMAINS = [
+    '.n8n.cloud',
+    '.n8n.io',
+    'api.dify.ai',
+    'cloud.dify.ai',
+    '.coze.com',
+    '.coze.cn',
+    'cloud.langflow.ai',
+    '.langflow.org',
+]
+
+LOCAL_PATTERNS = [
+    'localhost',
+    '127.0.0.1',
+    '0.0.0.0',
+    '192.168.',
+    '10.',
+    '172.16.',
+    '172.17.',
+    '172.18.',
+    '172.19.',
+    '172.20.',
+    '172.21.',
+    '172.22.',
+    '172.23.',
+    '172.24.',
+    '172.25.',
+    '172.26.',
+    '172.27.',
+    '172.28.',
+    '172.29.',
+    '172.30.',
+    '172.31.',
+]
+
+
+def get_runner_category(runner_name: str, runner_url: str) -> str:
+    if not runner_url:
+        return RunnerCategory.UNKNOWN
+
+    try:
+        parsed_url = urlparse(runner_url)
+        host = parsed_url.hostname.lower() if parsed_url.hostname else ''
+    except Exception:
+        return RunnerCategory.UNKNOWN
+
+    for pattern in LOCAL_PATTERNS:
+        if host.startswith(pattern):
+            return RunnerCategory.LOCAL
+
+    for domain in CLOUD_DOMAINS:
+        if host.endswith(domain):
+            return RunnerCategory.CLOUD
+
+    return RunnerCategory.CLOUD
+
+
+def get_runner_info(runner_name: str, runner_url: str) -> dict:
+    return {
+        'name': runner_name,
+        'url': runner_url,
+        'category': get_runner_category(runner_name, runner_url),
+    }
+
+
+def is_cloud_runner(runner_name: str, runner_url: str) -> bool:
+    return get_runner_category(runner_name, runner_url) == RunnerCategory.CLOUD
+
+
+def is_local_runner(runner_name: str, runner_url: str) -> bool:
+    return get_runner_category(runner_name, runner_url) == RunnerCategory.LOCAL
+
+
+def extract_runner_url(runner_name: str, runner, pipeline_config: dict | None) -> str | None:
+    if not runner or not hasattr(runner, 'pipeline_config'):
+        return None
+
+    ai_config = pipeline_config.get('ai', {}) if pipeline_config else {}
+
+    if runner_name == 'dify-service-api':
+        return ai_config.get('dify-service-api', {}).get('base-url')
+    elif runner_name == 'n8n-service-api':
+        return ai_config.get('n8n-service-api', {}).get('webhook-url')
+    elif runner_name == 'coze-api':
+        return ai_config.get('coze-api', {}).get('api-base')
+    elif runner_name == 'langflow-api':
+        return ai_config.get('langflow-api', {}).get('base-url')
+
+    return None
+
+
+def get_runner_category_from_runner(runner_name: str, runner, pipeline_config: dict | None) -> str:
+    runner_url = extract_runner_url(runner_name, runner, pipeline_config)
+    return get_runner_category(runner_name, runner_url)

src/langbot/pkg/vector/filter_utils.py

@@ -0,0 +1,69 @@
+"""Shared utilities for metadata filter handling across VDB backends.
+
+Canonical filter format (Chroma-style ``where`` syntax):
+
+    {"file_id": "abc"}                      # implicit $eq
+    {"file_id": {"$eq": "abc"}}             # explicit $eq
+    {"created_at": {"$gte": 1700000000}}    # comparison
+    {"file_type": {"$in": ["pdf", "docx"]}} # in-list
+
+Multiple top-level keys are AND-ed.  Supported operators:
+``$eq``, ``$ne``, ``$gt``, ``$gte``, ``$lt``, ``$lte``, ``$in``, ``$nin``.
+"""
+
+from __future__ import annotations
+
+import logging
+from typing import Any
+
+SUPPORTED_OPS = frozenset({'$eq', '$ne', '$gt', '$gte', '$lt', '$lte', '$in', '$nin'})
+
+logger = logging.getLogger(__name__)
+
+
+def normalize_filter(
+    raw: dict[str, Any] | None,
+) -> list[tuple[str, str, Any]]:
+    """Parse a canonical filter dict into ``[(field, op, value)]`` triples.
+
+    Returns an empty list when *raw* is ``None`` or empty.
+
+    Raises ``ValueError`` on unsupported operators or malformed entries.
+    """
+    if not raw:
+        return []
+
+    triples: list[tuple[str, str, Any]] = []
+    for field, condition in raw.items():
+        if isinstance(condition, dict):
+            for op, value in condition.items():
+                if op not in SUPPORTED_OPS:
+                    raise ValueError(f'Unsupported filter operator: {op}')
+                triples.append((field, op, value))
+        else:
+            # Bare value -> implicit $eq
+            triples.append((field, '$eq', condition))
+    return triples
+
+
+def strip_unsupported_fields(
+    triples: list[tuple[str, str, Any]],
+    supported_fields: set[str],
+) -> list[tuple[str, str, Any]]:
+    """Return only triples whose field is in *supported_fields*.
+
+    Dropped fields are logged at WARNING level so the caller knows they were
+    silently ignored (useful for Milvus / pgvector which only store a fixed
+    schema).
+    """
+    kept: list[tuple[str, str, Any]] = []
+    for field, op, value in triples:
+        if field in supported_fields:
+            kept.append((field, op, value))
+        else:
+            logger.warning(
+                'Filter field %r is not supported by this backend and will be ignored (supported: %s)',
+                field,
+                ', '.join(sorted(supported_fields)),
+            )
+    return kept

src/langbot/pkg/vector/mgr.py

@@ -1,7 +1,7 @@
 from __future__ import annotations
 
 from ..core import app
-from .vdb import VectorDatabase
+from .vdb import VectorDatabase, SearchType
 from .vdbs.chroma import ChromaVectorDatabase
 from .vdbs.qdrant import QdrantVectorDatabase
 from .vdbs.seekdb import SeekDBVectorDatabase
@@ -65,3 +65,95 @@ class VectorDBManager:
         else:
             self.vector_db = ChromaVectorDatabase(self.ap)
             self.ap.logger.warning('No vector database backend configured, defaulting to Chroma.')
+
+    def get_supported_search_types(self) -> list[str]:
+        """Return the search types supported by the current VDB backend."""
+        if self.vector_db is None:
+            return [SearchType.VECTOR.value]
+        return [st.value for st in self.vector_db.supported_search_types()]
+
+    async def upsert(
+        self,
+        collection_name: str,
+        vectors: list[list[float]],
+        ids: list[str],
+        metadata: list[dict] | None = None,
+        documents: list[str] | None = None,
+    ):
+        """Proxy: Upsert vectors"""
+        await self.vector_db.add_embeddings(
+            collection=collection_name,
+            ids=ids,
+            embeddings_list=vectors,
+            metadatas=metadata or [{} for _ in vectors],
+            documents=documents,
+        )
+
+    async def search(
+        self,
+        collection_name: str,
+        query_vector: list[float],
+        limit: int,
+        filter: dict | None = None,
+        search_type: str = 'vector',
+        query_text: str = '',
+    ) -> list[dict]:
+        """Proxy: Search vectors.
+
+        Returns a list of dicts with keys: 'id', 'distance', 'metadata'.
+        The underlying VectorDatabase.search returns Chroma-style format:
+        { 'ids': [['id1']], 'distances': [[0.1]], 'metadatas': [[{}]] }
+        """
+        results = await self.vector_db.search(
+            collection=collection_name,
+            query_embedding=query_vector,
+            k=limit,
+            search_type=search_type,
+            query_text=query_text,
+            filter=filter,
+        )
+
+        if not results or 'ids' not in results or not results['ids']:
+            return []
+
+        # Flatten nested lists (Chroma returns batch-style: list of lists)
+        raw_ids = results['ids']
+        raw_dists = results.get('distances', [])
+        raw_metas = results.get('metadatas', [])
+
+        r_ids = raw_ids[0] if raw_ids and isinstance(raw_ids[0], list) else raw_ids
+        r_dists = raw_dists[0] if raw_dists and isinstance(raw_dists[0], list) else raw_dists
+        r_metas = raw_metas[0] if raw_metas and isinstance(raw_metas[0], list) else raw_metas
+
+        parsed_results = []
+        for i, id_val in enumerate(r_ids):
+            parsed_results.append(
+                {
+                    'id': id_val,
+                    'distance': r_dists[i] if r_dists and i < len(r_dists) else 0.0,
+                    'metadata': r_metas[i] if r_metas and i < len(r_metas) else {},
+                }
+            )
+
+        return parsed_results
+
+    async def delete_by_file_id(self, collection_name: str, file_ids: list[str]):
+        """Proxy: Delete vectors by file_id (metadata-level identifier).
+
+        This delegates to VectorDatabase.delete_by_file_id which removes
+        all vectors associated with the given file IDs.
+        """
+        for file_id in file_ids:
+            await self.vector_db.delete_by_file_id(collection_name, file_id)
+
+    async def delete_collection(self, collection_name: str):
+        """Proxy: Delete an entire collection."""
+        await self.vector_db.delete_collection(collection_name)
+
+    async def delete_by_filter(self, collection_name: str, filter: dict) -> int:
+        """Proxy: Delete vectors by metadata filter.
+
+        Returns:
+            Number of deleted vectors (best-effort; some backends return 0).
+        """
+        return await self.vector_db.delete_by_filter(collection_name, filter)

src/langbot/pkg/vector/vdb.py

@@ -1,10 +1,28 @@
 from __future__ import annotations
 import abc
+import enum
 from typing import Any, Dict
 import numpy as np
 
 
+class SearchType(str, enum.Enum):
+    """Supported search types for vector databases."""
+
+    VECTOR = 'vector'
+    FULL_TEXT = 'full_text'
+    HYBRID = 'hybrid'
+
+
 class VectorDatabase(abc.ABC):
+    @classmethod
+    def supported_search_types(cls) -> list[SearchType]:
+        """Return the search types supported by this VDB backend.
+
+        Default: vector search only. Override in subclasses that support
+        full-text or hybrid search.
+        """
+        return [SearchType.VECTOR]
+
     @abc.abstractmethod
     async def add_embeddings(
         self,
@@ -12,14 +30,47 @@ class VectorDatabase(abc.ABC):
         ids: list[str],
         embeddings_list: list[list[float]],
         metadatas: list[dict[str, Any]],
-        documents: list[str],
+        documents: list[str] | None = None,
     ) -> None:
-        """Add vector data to the specified collection."""
+        """Add vector data to the specified collection.
+
+        Args:
+            collection: Collection name.
+            ids: Unique IDs for each vector.
+            embeddings_list: List of embedding vectors.
+            metadatas: List of metadata dicts.
+            documents: Optional raw text documents. Required for full-text
+                and hybrid search in backends that support them.
+        """
         pass
 
     @abc.abstractmethod
-    async def search(self, collection: str, query_embedding: np.ndarray, k: int = 5) -> Dict[str, Any]:
-        """Search for the most similar vectors in the specified collection."""
+    async def search(
+        self,
+        collection: str,
+        query_embedding: np.ndarray,
+        k: int = 5,
+        search_type: str = 'vector',
+        query_text: str = '',
+        filter: dict[str, Any] | None = None,
+    ) -> Dict[str, Any]:
+        """Search for the most similar vectors in the specified collection.
+
+        Args:
+            collection: Collection name.
+            query_embedding: Query vector for similarity search.
+            k: Number of results to return.
+            search_type: One of 'vector', 'full_text', 'hybrid'.
+            query_text: Raw query text, used for full_text and hybrid search.
+            filter: Optional metadata filters using Chroma-style ``where``
+                syntax.  Multiple top-level keys are AND-ed.  Supported
+                operators: ``$eq``, ``$ne``, ``$gt``, ``$gte``, ``$lt``,
+                ``$lte``, ``$in``, ``$nin``.  Example::
+
+                    {"file_id": "abc"}
+                    {"created_at": {"$gte": 1700000000}}
+                    {"file_type": {"$in": ["pdf", "docx"]}}
+        """
         pass
 
     @abc.abstractmethod
@@ -27,6 +78,20 @@ class VectorDatabase(abc.ABC):
         """Delete vectors from the specified collection by file_id."""
         pass
 
+    @abc.abstractmethod
+    async def delete_by_filter(self, collection: str, filter: dict[str, Any]) -> int:
+        """Delete vectors matching the given metadata filter.
+
+        Args:
+            collection: Collection name.
+            filter: Metadata filter dict in canonical format (see ``search``).
+
+        Returns:
+            Number of deleted vectors (best-effort; backends that cannot
+            report an exact count may return 0).
+        """
+        pass
+
     @abc.abstractmethod
     async def get_or_create_collection(self, collection: str):
         """Get or create collection."""

src/langbot/pkg/vector/vdbs/chroma.py

@@ -2,11 +2,14 @@ from __future__ import annotations
 import asyncio
 from typing import Any
 from chromadb import PersistentClient
-from langbot.pkg.vector.vdb import VectorDatabase
+from langbot.pkg.vector.vdb import VectorDatabase, SearchType
 from langbot.pkg.core import app
 import chromadb
 import chromadb.errors
 
+# RRF smoothing constant (standard value from the literature)
+_RRF_K = 60
+
 
 class ChromaVectorDatabase(VectorDatabase):
     def __init__(self, ap: app.Application, base_path: str = './data/chroma'):
@@ -14,6 +17,10 @@ class ChromaVectorDatabase(VectorDatabase):
         self.client = PersistentClient(path=base_path)
         self._collections = {}
 
+    @classmethod
+    def supported_search_types(cls) -> list[SearchType]:
+        return [SearchType.VECTOR, SearchType.FULL_TEXT, SearchType.HYBRID]
+
     async def get_or_create_collection(self, collection: str) -> chromadb.Collection:
         if collection not in self._collections:
             self._collections[collection] = await asyncio.to_thread(
@@ -28,27 +35,192 @@ class ChromaVectorDatabase(VectorDatabase):
         ids: list[str],
         embeddings_list: list[list[float]],
         metadatas: list[dict[str, Any]],
+        documents: list[str] | None = None,
     ) -> None:
         col = await self.get_or_create_collection(collection)
-        await asyncio.to_thread(col.add, embeddings=embeddings_list, ids=ids, metadatas=metadatas)
-        self.ap.logger.info(f"Added {len(ids)} embeddings to Chroma collection '{collection}'.")
+        kwargs: dict[str, Any] = dict(embeddings=embeddings_list, ids=ids, metadatas=metadatas)
+        if documents is not None:
+            kwargs['documents'] = documents
+        await asyncio.to_thread(col.upsert, **kwargs)
+        self.ap.logger.info(f"Upserted {len(ids)} embeddings to Chroma collection '{collection}'.")
 
-    async def search(self, collection: str, query_embedding: list[float], k: int = 5) -> dict[str, Any]:
+    async def search(
+        self,
+        collection: str,
+        query_embedding: list[float],
+        k: int = 5,
+        search_type: str = 'vector',
+        query_text: str = '',
+        filter: dict[str, Any] | None = None,
+    ) -> dict[str, Any]:
         col = await self.get_or_create_collection(collection)
-        results = await asyncio.to_thread(
-            col.query,
+
+        if search_type == SearchType.FULL_TEXT:
+            return await self._full_text_search(col, collection, k, query_text, filter)
+        elif search_type == SearchType.HYBRID:
+            return await self._hybrid_search(col, collection, query_embedding, k, query_text, filter)
+
+        # Default: vector search
+        return await self._vector_search(col, collection, query_embedding, k, filter)
+
+    async def _vector_search(
+        self,
+        col: chromadb.Collection,
+        collection: str,
+        query_embedding: list[float],
+        k: int,
+        filter: dict[str, Any] | None,
+    ) -> dict[str, Any]:
+        query_kwargs: dict[str, Any] = dict(
             query_embeddings=query_embedding,
             n_results=k,
             include=['metadatas', 'distances', 'documents'],
         )
-        self.ap.logger.info(f"Chroma search in '{collection}' returned {len(results.get('ids', [[]])[0])} results.")
+        if filter:
+            query_kwargs['where'] = filter
+        results = await asyncio.to_thread(col.query, **query_kwargs)
+        self.ap.logger.info(
+            f"Chroma vector search in '{collection}' returned {len(results.get('ids', [[]])[0])} results."
+        )
         return results
 
+    async def _full_text_search(
+        self,
+        col: chromadb.Collection,
+        collection: str,
+        k: int,
+        query_text: str,
+        filter: dict[str, Any] | None,
+    ) -> dict[str, Any]:
+        if not query_text:
+            return {'ids': [[]], 'metadatas': [[]], 'distances': [[]], 'documents': [[]]}
+
+        get_kwargs: dict[str, Any] = dict(
+            where_document={'$contains': query_text},
+            include=['metadatas', 'documents'],
+            limit=k,
+        )
+        if filter:
+            get_kwargs['where'] = filter
+        results = await asyncio.to_thread(col.get, **get_kwargs)
+
+        # col.get returns flat lists; wrap into column-major format.
+        # Distances are all 0.0 because Chroma's local $contains is a boolean
+        # filter with no relevance scoring.  Chroma's BM25 sparse embedding
+        # function (ChromaBm25EmbeddingFunction) can generate scored sparse
+        # vectors, but sparse vector *indexing* is only available on Chroma
+        # Cloud, not locally.  For ranked results, use hybrid mode or apply a
+        # reranker in a downstream stage.
+        ids = results.get('ids', [])
+        metadatas = results.get('metadatas', []) or [None] * len(ids)
+        documents = results.get('documents', []) or [None] * len(ids)
+        distances = [0.0] * len(ids)
+
+        self.ap.logger.info(f"Chroma full-text search in '{collection}' returned {len(ids)} results.")
+        return {'ids': [ids], 'metadatas': [metadatas], 'distances': [distances], 'documents': [documents]}
+
+    async def _hybrid_search(
+        self,
+        col: chromadb.Collection,
+        collection: str,
+        query_embedding: list[float],
+        k: int,
+        query_text: str,
+        filter: dict[str, Any] | None,
+    ) -> dict[str, Any]:
+        # Fall back to pure vector search when no text is provided
+        if not query_text:
+            return await self._vector_search(col, collection, query_embedding, k, filter)
+
+        # Run vector search and full-text search in parallel
+        vector_task = self._vector_search(col, collection, query_embedding, k, filter)
+        text_task = self._full_text_search(col, collection, k, query_text, filter)
+        vector_results, text_results = await asyncio.gather(vector_task, text_task)
+
+        vector_ids = vector_results.get('ids', [[]])[0]
+        text_ids = text_results.get('ids', [[]])[0]
+
+        if not vector_ids and not text_ids:
+            return {'ids': [[]], 'metadatas': [[]], 'distances': [[]], 'documents': [[]]}
+
+        # RRF fusion
+        fused = self._rrf_fuse([vector_ids, text_ids], k)
+        if not fused:
+            return {'ids': [[]], 'metadatas': [[]], 'distances': [[]], 'documents': [[]]}
+
+        fused_ids = [doc_id for doc_id, _ in fused]
+
+        # Fetch full metadata and documents for fused results
+        fetched = await asyncio.to_thread(col.get, ids=fused_ids, include=['metadatas', 'documents'])
+
+        # col.get returns results in arbitrary order; re-order to match fused ranking
+        fetched_map: dict[str, tuple] = {}
+        for i, fid in enumerate(fetched.get('ids', [])):
+            meta = (fetched.get('metadatas') or [None] * len(fetched['ids']))[i]
+            doc = (fetched.get('documents') or [None] * len(fetched['ids']))[i]
+            fetched_map[fid] = (meta, doc)
+
+        ordered_ids = []
+        ordered_metas = []
+        ordered_docs = []
+        ordered_dists = []
+
+        # Normalize RRF scores to 0~1 distances via min-max scaling.
+        # Raw RRF scores are tiny (e.g. 0.016~0.033 with k=60) so a naive
+        # ``1 - score`` would compress all distances into a narrow 0.96~0.98
+        # band with almost no discriminative power.  Min-max normalization
+        # spreads them across the full 0~1 range (0.0 = best match).
+        max_score = fused[0][1]
+        min_score = fused[-1][1]
+        score_range = max_score - min_score
+
+        for doc_id, score in fused:
+            if doc_id in fetched_map:
+                meta, doc = fetched_map[doc_id]
+                ordered_ids.append(doc_id)
+                ordered_metas.append(meta)
+                ordered_docs.append(doc)
+                if score_range > 0:
+                    ordered_dists.append(1.0 - (score - min_score) / score_range)
+                else:
+                    ordered_dists.append(0.0)
+
+        self.ap.logger.info(
+            f"Chroma hybrid search in '{collection}' returned {len(ordered_ids)} results "
+            f'(vector={len(vector_ids)}, text={len(text_ids)}).'
+        )
+        return {
+            'ids': [ordered_ids],
+            'metadatas': [ordered_metas],
+            'distances': [ordered_dists],
+            'documents': [ordered_docs],
+        }
+
+    @staticmethod
+    def _rrf_fuse(result_lists: list[list[str]], k: int) -> list[tuple[str, float]]:
+        """Reciprocal Rank Fusion over multiple ranked ID lists.
+
+        Returns a list of (doc_id, rrf_score) sorted by descending score,
+        truncated to *k* entries.
+        """
+        scores: dict[str, float] = {}
+        for ranked_ids in result_lists:
+            for rank, doc_id in enumerate(ranked_ids):
+                scores[doc_id] = scores.get(doc_id, 0.0) + 1.0 / (_RRF_K + rank + 1)
+        sorted_results = sorted(scores.items(), key=lambda x: x[1], reverse=True)
+        return sorted_results[:k]
+
     async def delete_by_file_id(self, collection: str, file_id: str) -> None:
         col = await self.get_or_create_collection(collection)
         await asyncio.to_thread(col.delete, where={'file_id': file_id})
         self.ap.logger.info(f"Deleted embeddings from Chroma collection '{collection}' with file_id: {file_id}")
 
+    async def delete_by_filter(self, collection: str, filter: dict[str, Any]) -> int:
+        col = await self.get_or_create_collection(collection)
+        await asyncio.to_thread(col.delete, where=filter)
+        self.ap.logger.info(f"Deleted embeddings from Chroma collection '{collection}' by filter")
+        return 0  # Chroma delete does not return a count
+
     async def delete_collection(self, collection: str):
         if collection in self._collections:
             del self._collections[collection]

src/langbot/pkg/vector/vdbs/milvus.py

@@ -4,8 +4,51 @@ from typing import Any, Dict
 from pymilvus import MilvusClient, DataType, CollectionSchema, FieldSchema
 from pymilvus.milvus_client.index import IndexParams
 from langbot.pkg.vector.vdb import VectorDatabase
+from langbot.pkg.vector.filter_utils import normalize_filter, strip_unsupported_fields
 from langbot.pkg.core import app
 
+# Milvus schema only stores these metadata fields; filter on other fields is
+# silently dropped with a warning.
+_MILVUS_SUPPORTED_FIELDS = {'text', 'file_id', 'chunk_uuid'}
+
+
+def _build_milvus_expr(filter_dict: dict[str, Any]) -> str:
+    """Translate canonical filter dict into a Milvus boolean expression string."""
+    triples = normalize_filter(filter_dict)
+    triples = strip_unsupported_fields(triples, _MILVUS_SUPPORTED_FIELDS)
+    if not triples:
+        return ''
+
+    parts: list[str] = []
+    for field, op, value in triples:
+        if op == '$eq':
+            parts.append(f'{field} == {_milvus_literal(value)}')
+        elif op == '$ne':
+            parts.append(f'{field} != {_milvus_literal(value)}')
+        elif op == '$gt':
+            parts.append(f'{field} > {_milvus_literal(value)}')
+        elif op == '$gte':
+            parts.append(f'{field} >= {_milvus_literal(value)}')
+        elif op == '$lt':
+            parts.append(f'{field} < {_milvus_literal(value)}')
+        elif op == '$lte':
+            parts.append(f'{field} <= {_milvus_literal(value)}')
+        elif op == '$in':
+            items = ', '.join(_milvus_literal(v) for v in value)
+            parts.append(f'{field} in [{items}]')
+        elif op == '$nin':
+            items = ', '.join(_milvus_literal(v) for v in value)
+            parts.append(f'{field} not in [{items}]')
+    return ' and '.join(parts)
+
+
+def _milvus_literal(value: Any) -> str:
+    """Format a Python value as a Milvus expression literal."""
+    if isinstance(value, str):
+        escaped = value.replace('\\', '\\\\').replace('"', '\\"')
+        return f'"{escaped}"'
+    return str(value)
+
 
 class MilvusVectorDatabase(VectorDatabase):
     """Milvus vector database implementation"""
@@ -155,6 +198,7 @@ class MilvusVectorDatabase(VectorDatabase):
         ids: list[str],
         embeddings_list: list[list[float]],
         metadatas: list[dict[str, Any]],
+        documents: list[str] | None = None,
     ) -> None:
         """Add vector embeddings to Milvus collection
 
@@ -200,7 +244,15 @@ class MilvusVectorDatabase(VectorDatabase):
 
         self.ap.logger.info(f"Added {len(ids)} embeddings to Milvus collection '{collection}'")
 
-    async def search(self, collection: str, query_embedding: list[float], k: int = 5) -> Dict[str, Any]:
+    async def search(
+        self,
+        collection: str,
+        query_embedding: list[float],
+        k: int = 5,
+        search_type: str = 'vector',
+        query_text: str = '',
+        filter: dict[str, Any] | None = None,
+    ) -> Dict[str, Any]:
         """Search for similar vectors in Milvus collection
 
         Args:
@@ -217,14 +269,19 @@ class MilvusVectorDatabase(VectorDatabase):
         # Perform search
         search_params = {'metric_type': 'COSINE', 'params': {}}
 
-        results = await asyncio.to_thread(
-            self.client.search,
+        search_kwargs: dict[str, Any] = dict(
             collection_name=collection,
             data=[query_embedding],
             limit=k,
             search_params=search_params,
             output_fields=['text', 'file_id', 'chunk_uuid'],
         )
+        if filter:
+            expr = _build_milvus_expr(filter)
+            if expr:
+                search_kwargs['filter'] = expr
+
+        results = await asyncio.to_thread(self.client.search, **search_kwargs)
 
         # Convert results to Chroma-compatible format
         # Milvus returns: [[ {id, distance, entity: {...}} ]]
@@ -268,6 +325,21 @@ class MilvusVectorDatabase(VectorDatabase):
         await asyncio.to_thread(self.client.delete, collection_name=collection, filter=f'file_id == "{file_id}"')
         self.ap.logger.info(f"Deleted embeddings from Milvus collection '{collection}' with file_id: {file_id}")
 
+    async def delete_by_filter(self, collection: str, filter: dict[str, Any]) -> int:
+        collection = self._normalize_collection_name(collection)
+        await self.get_or_create_collection(collection)
+
+        expr = _build_milvus_expr(filter)
+        if not expr:
+            self.ap.logger.warning(
+                f"Milvus delete_by_filter on '{collection}': filter produced empty expression, skipping"
+            )
+            return 0
+
+        await asyncio.to_thread(self.client.delete, collection_name=collection, filter=expr)
+        self.ap.logger.info(f"Deleted embeddings from Milvus collection '{collection}' by filter")
+        return 0  # Milvus delete does not return a count
+
     async def delete_collection(self, collection: str):
         """Delete a Milvus collection
 

src/langbot/pkg/vector/vdbs/pgvector_db.py

@@ -5,10 +5,21 @@ from sqlalchemy.orm import declarative_base
 from sqlalchemy.ext.asyncio import create_async_engine, AsyncSession, async_sessionmaker
 from pgvector.sqlalchemy import Vector
 from langbot.pkg.vector.vdb import VectorDatabase
+from langbot.pkg.vector.filter_utils import normalize_filter, strip_unsupported_fields
 from langbot.pkg.core import app
 
 Base = declarative_base()
 
+# pgvector schema only stores these metadata fields.
+_PG_SUPPORTED_FIELDS = {'text', 'file_id', 'chunk_uuid'}
+
+# Map schema field names to SQLAlchemy columns (resolved lazily from PgVectorEntry).
+_PG_COLUMN_MAP = {
+    'text': 'text',
+    'file_id': 'file_id',
+    'chunk_uuid': 'chunk_uuid',
+}
+
 
 class PgVectorEntry(Base):
     """SQLAlchemy model for pgvector entries"""
@@ -23,6 +34,33 @@ class PgVectorEntry(Base):
     chunk_uuid = Column(String)
 
 
+def _build_pg_conditions(filter_dict: dict[str, Any]) -> list:
+    """Translate canonical filter dict into a list of SQLAlchemy conditions."""
+    triples = normalize_filter(filter_dict)
+    triples = strip_unsupported_fields(triples, _PG_SUPPORTED_FIELDS)
+
+    conditions = []
+    for field, op, value in triples:
+        col = getattr(PgVectorEntry, _PG_COLUMN_MAP[field])
+        if op == '$eq':
+            conditions.append(col == value)
+        elif op == '$ne':
+            conditions.append(col != value)
+        elif op == '$gt':
+            conditions.append(col > value)
+        elif op == '$gte':
+            conditions.append(col >= value)
+        elif op == '$lt':
+            conditions.append(col < value)
+        elif op == '$lte':
+            conditions.append(col <= value)
+        elif op == '$in':
+            conditions.append(col.in_(value))
+        elif op == '$nin':
+            conditions.append(col.notin_(value))
+    return conditions
+
+
 class PgVectorDatabase(VectorDatabase):
     """PostgreSQL with pgvector extension database implementation"""
 
@@ -109,6 +147,7 @@ class PgVectorDatabase(VectorDatabase):
         ids: list[str],
         embeddings_list: list[list[float]],
         metadatas: list[dict[str, Any]],
+        documents: list[str] | None = None,
     ) -> None:
         """Add vector embeddings to pgvector
 
@@ -142,7 +181,15 @@ class PgVectorDatabase(VectorDatabase):
                 self.ap.logger.error(f'Error adding embeddings to pgvector: {e}')
                 raise
 
-    async def search(self, collection: str, query_embedding: list[float], k: int = 5) -> Dict[str, Any]:
+    async def search(
+        self,
+        collection: str,
+        query_embedding: list[float],
+        k: int = 5,
+        search_type: str = 'vector',
+        query_text: str = '',
+        filter: dict[str, Any] | None = None,
+    ) -> Dict[str, Any]:
         """Search for similar vectors using cosine distance
 
         Args:
@@ -174,6 +221,10 @@ class PgVectorDatabase(VectorDatabase):
                     .limit(k)
                 )
 
+                if filter:
+                    for cond in _build_pg_conditions(filter):
+                        stmt = stmt.filter(cond)
+
                 result = await session.execute(stmt)
                 rows = result.fetchall()
 
@@ -225,6 +276,39 @@ class PgVectorDatabase(VectorDatabase):
                 self.ap.logger.error(f'Error deleting from pgvector: {e}')
                 raise
 
+    async def delete_by_filter(self, collection: str, filter: dict[str, Any]) -> int:
+        """Delete vectors matching a metadata filter.
+
+        Args:
+            collection: Collection name
+            filter: Canonical metadata filter dict
+        """
+        conditions = _build_pg_conditions(filter)
+        if not conditions:
+            self.ap.logger.warning(
+                f"pgvector delete_by_filter on '{collection}': filter produced no conditions, skipping"
+            )
+            return 0
+
+        await self.get_or_create_collection(collection)
+
+        async with self.AsyncSessionLocal() as session:
+            try:
+                from sqlalchemy import delete
+
+                stmt = delete(PgVectorEntry).where(PgVectorEntry.collection == collection)
+                for cond in conditions:
+                    stmt = stmt.where(cond)
+                result = await session.execute(stmt)
+                await session.commit()
+                deleted = result.rowcount
+                self.ap.logger.info(f"Deleted {deleted} embeddings from pgvector collection '{collection}' by filter")
+                return deleted
+            except Exception as e:
+                await session.rollback()
+                self.ap.logger.error(f'Error deleting from pgvector by filter: {e}')
+                raise
+
     async def delete_collection(self, collection: str):
         """Delete all vectors in a collection
 

src/langbot/pkg/vector/vdbs/qdrant.py

@@ -5,6 +5,37 @@ from typing import Any, Dict, List
 from qdrant_client import AsyncQdrantClient, models
 from langbot.pkg.core import app
 from langbot.pkg.vector.vdb import VectorDatabase
+from langbot.pkg.vector.filter_utils import normalize_filter
+
+
+def _build_qdrant_filter(filter_dict: dict[str, Any]) -> models.Filter:
+    """Translate canonical filter dict into a Qdrant ``models.Filter``."""
+    triples = normalize_filter(filter_dict)
+    must: list[models.Condition] = []
+    must_not: list[models.Condition] = []
+
+    for field, op, value in triples:
+        if op == '$eq':
+            must.append(models.FieldCondition(key=field, match=models.MatchValue(value=value)))
+        elif op == '$ne':
+            must_not.append(models.FieldCondition(key=field, match=models.MatchValue(value=value)))
+        elif op == '$in':
+            must.append(models.FieldCondition(key=field, match=models.MatchAny(any=value)))
+        elif op == '$nin':
+            must_not.append(models.FieldCondition(key=field, match=models.MatchAny(any=value)))
+        elif op in ('$gt', '$gte', '$lt', '$lte'):
+            range_kwargs: dict[str, Any] = {}
+            if op == '$gt':
+                range_kwargs['gt'] = value
+            elif op == '$gte':
+                range_kwargs['gte'] = value
+            elif op == '$lt':
+                range_kwargs['lt'] = value
+            elif op == '$lte':
+                range_kwargs['lte'] = value
+            must.append(models.FieldCondition(key=field, range=models.Range(**range_kwargs)))
+
+    return models.Filter(must=must or None, must_not=must_not or None)
 
 
 class QdrantVectorDatabase(VectorDatabase):
@@ -48,6 +79,7 @@ class QdrantVectorDatabase(VectorDatabase):
         ids: List[str],
         embeddings_list: List[List[float]],
         metadatas: List[Dict[str, Any]],
+        documents: List[str] | None = None,
     ) -> None:
         if not embeddings_list:
             return
@@ -60,19 +92,29 @@ class QdrantVectorDatabase(VectorDatabase):
         await self.client.upsert(collection_name=collection, points=points)
         self.ap.logger.info(f"Added {len(ids)} embeddings to Qdrant collection '{collection}'.")
 
-    async def search(self, collection: str, query_embedding: list[float], k: int = 5) -> dict[str, Any]:
+    async def search(
+        self,
+        collection: str,
+        query_embedding: list[float],
+        k: int = 5,
+        search_type: str = 'vector',
+        query_text: str = '',
+        filter: dict[str, Any] | None = None,
+    ) -> dict[str, Any]:
         exists = await self.client.collection_exists(collection)
         if not exists:
             return {'ids': [[]], 'metadatas': [[]], 'distances': [[]]}
 
-        hits = (
-            await self.client.query_points(
-                collection_name=collection,
-                query=query_embedding,
-                limit=k,
-                with_payload=True,
-            )
-        ).points
+        query_kwargs: dict[str, Any] = dict(
+            collection_name=collection,
+            query=query_embedding,
+            limit=k,
+            with_payload=True,
+        )
+        if filter:
+            query_kwargs['query_filter'] = _build_qdrant_filter(filter)
+
+        hits = (await self.client.query_points(**query_kwargs)).points
         ids = [str(hit.id) for hit in hits]
         metadatas = [hit.payload or {} for hit in hits]
         # Qdrant's score is similarity; convert to a pseudo-distance for consistency
@@ -95,6 +137,19 @@ class QdrantVectorDatabase(VectorDatabase):
         )
         self.ap.logger.info(f"Deleted embeddings from Qdrant collection '{collection}' with file_id: {file_id}")
 
+    async def delete_by_filter(self, collection: str, filter: dict[str, Any]) -> int:
+        exists = await self.client.collection_exists(collection)
+        if not exists:
+            return 0
+
+        qdrant_filter = _build_qdrant_filter(filter)
+        await self.client.delete(
+            collection_name=collection,
+            points_selector=qdrant_filter,
+        )
+        self.ap.logger.info(f"Deleted embeddings from Qdrant collection '{collection}' by filter")
+        return 0  # Qdrant delete does not return a count
+
     async def delete_collection(self, collection: str):
         try:
             await self.client.delete_collection(collection)

src/langbot/pkg/vector/vdbs/seekdb.py

@@ -5,7 +5,7 @@ from typing import Any, Dict, List
 
 
 from langbot.pkg.core import app
-from langbot.pkg.vector.vdb import VectorDatabase
+from langbot.pkg.vector.vdb import VectorDatabase, SearchType
 
 try:
     import pyseekdb
@@ -25,9 +25,13 @@ class SeekDBVectorDatabase(VectorDatabase):
     SeekDB is an AI-native search database by OceanBase that unifies
     relational, vector, text, JSON and GIS in a single engine.
 
-    Supports both embedded mode and remote server mode.
+    Supports embedded mode, remote server mode, and full-text/hybrid search.
     """
 
+    @classmethod
+    def supported_search_types(cls) -> list[SearchType]:
+        return [SearchType.VECTOR, SearchType.FULL_TEXT, SearchType.HYBRID]
+
     def __init__(self, ap: app.Application):
         if not SEEKDB_AVAILABLE:
             raise ImportError('pyseekdb is not installed. Install it with: pip install pyseekdb')
@@ -89,6 +93,7 @@ class SeekDBVectorDatabase(VectorDatabase):
             {
                 '\x00': '',
                 '\\': '\\\\',
+                "'": "''",  # Standard SQL escaping (OceanBase NO_BACKSLASH_ESCAPES)
                 '"': '\\"',
                 '\n': '\\n',
                 '\r': '\\r',
@@ -111,8 +116,10 @@ class SeekDBVectorDatabase(VectorDatabase):
 
         # Collection doesn't exist, create it
         if vector_size is None:
-            # Default dimension if not specified
-            vector_size = 384
+            raise ValueError(
+                f"Cannot create SeekDB collection '{collection}' without knowing the vector dimension. "
+                'Ensure add_embeddings is called before any standalone get_or_create_collection.'
+            )
 
         # Create HNSW configuration
         config = HNSWConfiguration(dimension=vector_size, distance='cosine')
@@ -147,7 +154,12 @@ class SeekDBVectorDatabase(VectorDatabase):
         return await self._get_or_create_collection_internal(collection)
 
     async def add_embeddings(
-        self, collection: str, ids: List[str], embeddings_list: List[List[float]], metadatas: List[Dict[str, Any]]
+        self,
+        collection: str,
+        ids: List[str],
+        embeddings_list: List[List[float]],
+        metadatas: List[Dict[str, Any]],
+        documents: List[str] | None = None,
     ) -> None:
         """Add vector embeddings to the specified collection.
 
@@ -156,6 +168,7 @@ class SeekDBVectorDatabase(VectorDatabase):
             ids: List of document IDs
             embeddings_list: List of embedding vectors
             metadatas: List of metadata dictionaries
+            documents: Optional raw text documents for full-text search support
         """
         if not embeddings_list:
             return
@@ -166,17 +179,33 @@ class SeekDBVectorDatabase(VectorDatabase):
 
         cleaned_metadatas = [self._clean_metadata(meta) for meta in metadatas]
 
-        await asyncio.to_thread(coll.add, ids=ids, embeddings=embeddings_list, metadatas=cleaned_metadatas)
+        kwargs: Dict[str, Any] = dict(ids=ids, embeddings=embeddings_list, metadatas=cleaned_metadatas)
+        if documents is not None:
+            kwargs['documents'] = [doc.translate(self._escape_table) for doc in documents]
+        await asyncio.to_thread(coll.add, **kwargs)
 
         self.ap.logger.info(f"Added {len(ids)} embeddings to SeekDB collection '{collection}'")
 
-    async def search(self, collection: str, query_embedding: List[float], k: int = 5) -> Dict[str, Any]:
+    async def search(
+        self,
+        collection: str,
+        query_embedding: List[float],
+        k: int = 5,
+        search_type: str = 'vector',
+        query_text: str = '',
+        filter: Dict[str, Any] | None = None,
+    ) -> Dict[str, Any]:
         """Search for the most similar vectors in the specified collection.
 
+        SeekDB supports vector, full-text, and hybrid search modes.
+
         Args:
             collection: Collection name
-            query_embedding: Query vector
+            query_embedding: Query vector (used for vector and hybrid modes)
             k: Number of results to return
+            search_type: One of 'vector', 'full_text', 'hybrid'
+            query_text: Raw query text (used for full_text and hybrid modes)
+            filter: Optional metadata filters (Chroma-style ``where`` syntax).
 
         Returns:
             Dictionary with 'ids', 'metadatas', 'distances' keys
@@ -193,11 +222,73 @@ class SeekDBVectorDatabase(VectorDatabase):
         else:
             coll = self._collections[collection]
 
-        # Perform query
-        # SeekDB's query() returns: {'ids': [[...]], 'metadatas': [[...]], 'distances': [[...]]}
-        results = await asyncio.to_thread(coll.query, query_embeddings=query_embedding, n_results=k)
+        # Route by search type.
+        # pyseekdb's query() always requires embeddings, so full-text and
+        # hybrid modes use hybrid_search() which supports text-only queries
+        # and returns the same nested-list format with distances.
+        if search_type == SearchType.FULL_TEXT:
+            if not query_text:
+                return {'ids': [[]], 'metadatas': [[]], 'distances': [[]]}
 
-        self.ap.logger.info(f"SeekDB search in '{collection}' returned {len(results.get('ids', [[]])[0])} results")
+            query_cfg: Dict[str, Any] = {
+                'where_document': {'$contains': query_text},
+                'n_results': k,
+            }
+            if filter:
+                query_cfg['where'] = filter
+
+            # TODO: pyseekdb hybrid_search with query-only (no knn) returns None
+            # for IDs due to column name mismatch (*/_id vs _id).
+            # See: https://github.com/oceanbase/pyseekdb/issues/171
+            results = await asyncio.to_thread(
+                coll.hybrid_search,
+                query=query_cfg,
+                knn=None,
+                n_results=k,
+                include=['documents', 'metadatas'],
+            )
+
+        elif search_type == SearchType.HYBRID:
+            if not query_text:
+                # Fall back to pure vector search when no text is provided
+                query_kwargs: Dict[str, Any] = {
+                    'n_results': k,
+                    'query_embeddings': query_embedding,
+                }
+                if filter:
+                    query_kwargs['where'] = filter
+                results = await asyncio.to_thread(coll.query, **query_kwargs)
+            else:
+                query_cfg = {
+                    'where_document': {'$contains': query_text},
+                    'n_results': k,
+                }
+                knn_cfg: Dict[str, Any] = {
+                    'query_embeddings': query_embedding,
+                    'n_results': k,
+                }
+                if filter:
+                    query_cfg['where'] = filter
+                    knn_cfg['where'] = filter
+
+                results = await asyncio.to_thread(
+                    coll.hybrid_search,
+                    query=query_cfg,
+                    knn=knn_cfg,
+                    rank={'rrf': {}},
+                    n_results=k,
+                    include=['documents', 'metadatas'],
+                )
+        else:
+            # Default: vector search via query()
+            query_kwargs = {'n_results': k, 'query_embeddings': query_embedding}
+            if filter:
+                query_kwargs['where'] = filter
+            results = await asyncio.to_thread(coll.query, **query_kwargs)
+
+        self.ap.logger.info(
+            f"SeekDB {search_type} search in '{collection}' returned {len(results.get('ids', [[]])[0])} results"
+        )
 
         return results
 
@@ -227,6 +318,28 @@ class SeekDBVectorDatabase(VectorDatabase):
 
         self.ap.logger.info(f"Deleted embeddings from SeekDB collection '{collection}' with file_id: {file_id}")
 
+    async def delete_by_filter(self, collection: str, filter: Dict[str, Any]) -> int:
+        """Delete vectors from the collection by metadata filter.
+
+        Args:
+            collection: Collection name
+            filter: Chroma-style ``where`` filter dict
+        """
+        exists = await asyncio.to_thread(self.client.has_collection, collection)
+        if not exists:
+            self.ap.logger.warning(f"SeekDB collection '{collection}' not found for deletion")
+            return 0
+
+        if collection not in self._collections:
+            coll = await asyncio.to_thread(self.client.get_collection, collection, embedding_function=None)
+            self._collections[collection] = coll
+        else:
+            coll = self._collections[collection]
+
+        await asyncio.to_thread(coll.delete, where=filter)
+        self.ap.logger.info(f"Deleted embeddings from SeekDB collection '{collection}' by filter")
+        return 0  # SeekDB delete does not return a count
+
     async def delete_collection(self, collection: str):
         """Delete the entire collection.
 

src/langbot/templates/config.yaml

@@ -2,6 +2,7 @@ admins: []
 api:
     port: 5300
     webhook_prefix: 'http://127.0.0.1:5300'
+    extra_webhook_prefix: ''
 command:
     enable: true
     prefix:
@@ -15,6 +16,7 @@ proxy:
     http: ''
     https: ''
 system:
+    instance_id: ''
     edition: community
     recovery_key: ''
     allow_modify_login_info: true

src/langbot/templates/default-pipeline-config.json

@@ -41,7 +41,10 @@
             "runner": "local-agent"
         },
         "local-agent": {
-            "model": "",
+            "model": {
+                "primary": "",
+                "fallbacks": []
+            },
             "max-round": 10,
             "prompt": [
                 {
@@ -95,11 +98,12 @@
             "max": 0
         },
         "misc": {
-            "hide-exception": true,
+            "exception-handling": "show-hint",
+            "failure-hint": "Request failed.",
             "at-sender": true,
             "quote-origin": true,
             "track-function-calls": false,
             "remove-think": false
         }
     }
-}
+}

src/langbot/templates/metadata/pipeline/ai.yaml

@@ -59,8 +59,11 @@ stages:
         label:
           en_US: Model
           zh_Hans: 模型
-        type: llm-model-selector
+        type: model-fallback-selector
         required: true
+        default:
+          primary: ''
+          fallbacks: []
       - name: max-round
         label:
           en_US: Max Round

src/langbot/templates/metadata/pipeline/output.yaml

@@ -78,13 +78,39 @@ stages:
       en_US: Misc
       zh_Hans: 杂项
     config:
-      - name: hide-exception
+      - name: exception-handling
         label:
-          en_US: Hide Exception
-          zh_Hans: 不输出异常信息给用户
-        type: boolean
+          en_US: Exception Handling Strategy
+          zh_Hans: 异常处理策略
+        description:
+          en_US: Controls how error messages are displayed to the user when an AI request fails
+          zh_Hans: 控制 AI 请求失败时向用户展示错误信息的方式
+        type: select
         required: true
-        default: true
+        default: show-hint
+        options:
+          - name: show-error
+            label:
+              en_US: Show Full Error
+              zh_Hans: 显示完整报错信息
+          - name: show-hint
+            label:
+              en_US: Show Failure Hint
+              zh_Hans: 仅文字提示
+          - name: hide
+            label:
+              en_US: Hide All
+              zh_Hans: 不显示任何异常信息
+      - name: failure-hint
+        label:
+          en_US: Failure Hint Text
+          zh_Hans: 失败提示文本
+        description:
+          en_US: The text to display when a request fails. Only effective when Exception Handling Strategy is set to "Show Failure Hint"
+          zh_Hans: 请求失败时显示的提示文本，仅在异常处理策略设置为"仅文字提示"时生效
+        type: string
+        required: false
+        default: 'Request failed.'
       - name: at-sender
         label:
           en_US: At Sender
@@ -119,3 +145,4 @@ stages:
         type: boolean
         required: true
         default: false
+

tests/unit_tests/config/test_webhook_display_prefix.py

@@ -91,14 +91,15 @@ class TestWebhookDisplayPrefix:
 
     def test_default_webhook_prefix(self):
         """Test that the default webhook display prefix is correctly set"""
-        cfg = {'api': {'port': 5300, 'webhook_prefix': 'http://127.0.0.1:5300'}}
+        cfg = {'api': {'port': 5300, 'webhook_prefix': 'http://127.0.0.1:5300', 'extra_webhook_prefix': ''}}
 
         # Should have the default value
         assert cfg['api']['webhook_prefix'] == 'http://127.0.0.1:5300'
+        assert cfg['api']['extra_webhook_prefix'] == ''
 
     def test_webhook_prefix_env_override(self):
         """Test overriding webhook_prefix via environment variable"""
-        cfg = {'api': {'port': 5300, 'webhook_prefix': 'http://127.0.0.1:5300'}}
+        cfg = {'api': {'port': 5300, 'webhook_prefix': 'http://127.0.0.1:5300', 'extra_webhook_prefix': ''}}
 
         # Set environment variable
         os.environ['API__WEBHOOK_PREFIX'] = 'https://example.com:8080'
@@ -112,7 +113,7 @@ class TestWebhookDisplayPrefix:
 
     def test_webhook_prefix_with_custom_domain(self):
         """Test webhook_prefix with custom domain"""
-        cfg = {'api': {'port': 5300, 'webhook_prefix': 'http://127.0.0.1:5300'}}
+        cfg = {'api': {'port': 5300, 'webhook_prefix': 'http://127.0.0.1:5300', 'extra_webhook_prefix': ''}}
 
         # Set to a custom domain
         os.environ['API__WEBHOOK_PREFIX'] = 'https://bot.mycompany.com'
@@ -126,7 +127,7 @@ class TestWebhookDisplayPrefix:
 
     def test_webhook_prefix_with_subdirectory(self):
         """Test webhook_prefix with subdirectory path"""
-        cfg = {'api': {'port': 5300, 'webhook_prefix': 'http://127.0.0.1:5300'}}
+        cfg = {'api': {'port': 5300, 'webhook_prefix': 'http://127.0.0.1:5300', 'extra_webhook_prefix': ''}}
 
         # Set to a URL with subdirectory
         os.environ['API__WEBHOOK_PREFIX'] = 'https://example.com/langbot'
@@ -138,6 +139,37 @@ class TestWebhookDisplayPrefix:
         # Cleanup
         del os.environ['API__WEBHOOK_PREFIX']
 
+    def test_extra_webhook_prefix_default_empty(self):
+        """Test that extra_webhook_prefix defaults to empty string"""
+        cfg = {'api': {'port': 5300, 'webhook_prefix': 'http://127.0.0.1:5300', 'extra_webhook_prefix': ''}}
+
+        bot_uuid = 'test-bot-uuid'
+        webhook_prefix = cfg['api'].get('webhook_prefix', 'http://127.0.0.1:5300')
+        extra_webhook_prefix = cfg['api'].get('extra_webhook_prefix', '')
+        webhook_url = f'/bots/{bot_uuid}'
+
+        assert f'{webhook_prefix}{webhook_url}' == 'http://127.0.0.1:5300/bots/test-bot-uuid'
+        # extra should be empty when not configured
+        assert extra_webhook_prefix == ''
+
+    def test_extra_webhook_prefix_env_override(self):
+        """Test overriding extra_webhook_prefix via environment variable"""
+        cfg = {'api': {'port': 5300, 'webhook_prefix': 'http://127.0.0.1:5300', 'extra_webhook_prefix': ''}}
+
+        os.environ['API__EXTRA_WEBHOOK_PREFIX'] = 'https://extra.example.com'
+
+        result = _apply_env_overrides_to_config(cfg)
+
+        assert result['api']['extra_webhook_prefix'] == 'https://extra.example.com'
+
+        bot_uuid = 'test-bot-uuid'
+        extra_prefix = result['api']['extra_webhook_prefix']
+        webhook_url = f'/bots/{bot_uuid}'
+        assert f'{extra_prefix}{webhook_url}' == 'https://extra.example.com/bots/test-bot-uuid'
+
+        # Cleanup
+        del os.environ['API__EXTRA_WEBHOOK_PREFIX']
+
 
 if __name__ == '__main__':
     pytest.main([__file__, '-v'])

tests/unit_tests/pipeline/conftest.py

@@ -194,7 +194,7 @@ def sample_query(sample_message_chain, sample_message_event, mock_adapter):
         pipeline_config={
             'ai': {
                 'runner': {'runner': 'local-agent'},
-                'local-agent': {'model': 'test-model-uuid', 'prompt': 'test-prompt'},
+                'local-agent': {'model': {'primary': 'test-model-uuid', 'fallbacks': []}, 'prompt': 'test-prompt'},
             },
             'output': {'misc': {'at-sender': False, 'quote-origin': False}},
             'trigger': {'misc': {'combine-quote-message': False}},
@@ -219,7 +219,7 @@ def sample_pipeline_config():
     return {
         'ai': {
             'runner': {'runner': 'local-agent'},
-            'local-agent': {'model': 'test-model-uuid', 'prompt': 'test-prompt'},
+            'local-agent': {'model': {'primary': 'test-model-uuid', 'fallbacks': []}, 'prompt': 'test-prompt'},
         },
         'output': {'misc': {'at-sender': False, 'quote-origin': False}},
         'trigger': {'misc': {'combine-quote-message': False}},

tests/unit_tests/pipeline/test_config_coercion.py

@@ -0,0 +1,113 @@
+"""Unit tests for config_coercion module"""
+
+from __future__ import annotations
+
+import pytest
+
+from langbot.pkg.pipeline.config_coercion import _coerce_value, coerce_pipeline_config
+
+
+class TestCoerceValue:
+    """Tests for _coerce_value function"""
+
+    def test_none_passthrough(self):
+        assert _coerce_value(None, 'integer') is None
+        assert _coerce_value(None, 'boolean') is None
+
+    def test_string_to_integer(self):
+        assert _coerce_value('120', 'integer') == 120
+        assert _coerce_value('0', 'integer') == 0
+        assert _coerce_value('-5', 'integer') == -5
+
+    def test_integer_passthrough(self):
+        assert _coerce_value(42, 'integer') == 42
+
+    def test_string_to_float(self):
+        assert _coerce_value('3.14', 'number') == 3.14
+        assert _coerce_value('3.14', 'float') == 3.14
+
+    def test_int_to_float(self):
+        assert _coerce_value(3, 'number') == 3.0
+        assert isinstance(_coerce_value(3, 'number'), float)
+
+    def test_float_passthrough(self):
+        assert _coerce_value(3.14, 'float') == 3.14
+
+    def test_string_to_bool(self):
+        assert _coerce_value('true', 'boolean') is True
+        assert _coerce_value('True', 'boolean') is True
+        assert _coerce_value('false', 'boolean') is False
+        assert _coerce_value('False', 'boolean') is False
+
+    def test_bool_passthrough(self):
+        assert _coerce_value(True, 'boolean') is True
+        assert _coerce_value(False, 'boolean') is False
+
+    def test_invalid_bool_string_raises(self):
+        with pytest.raises(ValueError):
+            _coerce_value('notabool', 'boolean')
+
+    def test_unknown_type_passthrough(self):
+        assert _coerce_value('hello', 'string') == 'hello'
+        assert _coerce_value('hello', 'unknown') == 'hello'
+
+    def test_invalid_integer_raises(self):
+        with pytest.raises(ValueError):
+            _coerce_value('abc', 'integer')
+
+
+class TestCoercePipelineConfig:
+    """Tests for coerce_pipeline_config function"""
+
+    def _make_meta(self, section_name: str, stage_name: str, fields: list[dict]) -> dict:
+        return {
+            'name': section_name,
+            'stages': [{'name': stage_name, 'config': fields}],
+        }
+
+    def test_coerce_integer_in_config(self):
+        config = {'trigger': {'misc': {'timeout': '120'}}}
+        meta = self._make_meta('trigger', 'misc', [{'name': 'timeout', 'type': 'integer'}])
+        coerce_pipeline_config(config, meta)
+        assert config['trigger']['misc']['timeout'] == 120
+
+    def test_coerce_boolean_in_config(self):
+        config = {'output': {'misc': {'at-sender': 'true'}}}
+        meta = self._make_meta('output', 'misc', [{'name': 'at-sender', 'type': 'boolean'}])
+        coerce_pipeline_config(config, meta)
+        assert config['output']['misc']['at-sender'] is True
+
+    def test_missing_section_skipped(self):
+        config = {'ai': {}}
+        meta = self._make_meta('trigger', 'misc', [{'name': 'x', 'type': 'integer'}])
+        coerce_pipeline_config(config, meta)  # should not raise
+
+    def test_missing_field_skipped(self):
+        config = {'trigger': {'misc': {}}}
+        meta = self._make_meta('trigger', 'misc', [{'name': 'nonexistent', 'type': 'integer'}])
+        coerce_pipeline_config(config, meta)  # should not raise
+
+    def test_invalid_value_logs_warning(self, caplog):
+        config = {'trigger': {'misc': {'timeout': 'abc'}}}
+        meta = self._make_meta('trigger', 'misc', [{'name': 'timeout', 'type': 'integer'}])
+        import logging
+
+        with caplog.at_level(logging.WARNING):
+            coerce_pipeline_config(config, meta)
+        assert config['trigger']['misc']['timeout'] == 'abc'  # unchanged
+        assert 'Failed to coerce' in caplog.text
+
+    def test_empty_metadata(self):
+        config = {'trigger': {'misc': {'timeout': '120'}}}
+        coerce_pipeline_config(config)  # no metadata args, should not raise
+
+    def test_multiple_metadata(self):
+        config = {
+            'trigger': {'misc': {'timeout': '120'}},
+            'output': {'misc': {'at-sender': 'false'}},
+        }
+        meta_trigger = self._make_meta('trigger', 'misc', [{'name': 'timeout', 'type': 'integer'}])
+        meta_output = self._make_meta('output', 'misc', [{'name': 'at-sender', 'type': 'boolean'}])
+        coerce_pipeline_config(config, meta_trigger, meta_output)
+        assert config['trigger']['misc']['timeout'] == 120
+        assert config['output']['misc']['at-sender'] is False

tests/unit_tests/plugin/test_plugin_component_filtering.py

@@ -38,13 +38,11 @@ async def test_plugin_list_filter_by_component_kinds():
                 'manifest': {
                     'metadata': {
                         'author': 'author2',
-                        'name': 'plugin_with_knowledge_retriever_only',
+                        'name': 'plugin_with_knowledge_engine_only',
                     }
                 }
             },
-            'components': [
-                {'manifest': {'manifest': {'kind': 'KnowledgeRetriever', 'metadata': {'name': 'retriever1'}}}}
-            ],
+            'components': [{'manifest': {'manifest': {'kind': 'KnowledgeEngine', 'metadata': {'name': 'retriever1'}}}}],
         },
         {
             'debug': False,
@@ -81,7 +79,7 @@ async def test_plugin_list_filter_by_component_kinds():
                 }
             },
             'components': [
-                {'manifest': {'manifest': {'kind': 'KnowledgeRetriever', 'metadata': {'name': 'retriever2'}}}},
+                {'manifest': {'manifest': {'kind': 'KnowledgeEngine', 'metadata': {'name': 'retriever2'}}}},
                 {'manifest': {'manifest': {'kind': 'Tool', 'metadata': {'name': 'tool2'}}}},
             ],
         },
@@ -108,8 +106,8 @@ async def test_plugin_list_filter_by_component_kinds():
     assert 'plugin_with_command' in plugin_names
     assert 'plugin_with_event_listener' in plugin_names
     assert 'plugin_with_mixed_components' in plugin_names
-    # Plugin with only KnowledgeRetriever should NOT be included
-    assert 'plugin_with_knowledge_retriever_only' not in plugin_names
+    # Plugin with only KnowledgeEngine should NOT be included
+    assert 'plugin_with_knowledge_engine_only' not in plugin_names
 
 
 @pytest.mark.asyncio
@@ -150,9 +148,7 @@ async def test_plugin_list_filter_no_filter():
                     }
                 }
             },
-            'components': [
-                {'manifest': {'manifest': {'kind': 'KnowledgeRetriever', 'metadata': {'name': 'retriever1'}}}}
-            ],
+            'components': [{'manifest': {'manifest': {'kind': 'KnowledgeEngine', 'metadata': {'name': 'retriever1'}}}}],
         },
     ]
 
@@ -189,7 +185,7 @@ async def test_plugin_list_filter_empty_result():
     connector = PluginRuntimeConnector(mock_app, AsyncMock())
     connector.handler = MagicMock()
 
-    # Mock plugin data - only KnowledgeRetriever plugins
+    # Mock plugin data - only KnowledgeEngine plugins
     mock_plugins = [
         {
             'debug': False,
@@ -201,9 +197,7 @@ async def test_plugin_list_filter_empty_result():
                     }
                 }
             },
-            'components': [
-                {'manifest': {'manifest': {'kind': 'KnowledgeRetriever', 'metadata': {'name': 'retriever1'}}}}
-            ],
+            'components': [{'manifest': {'manifest': {'kind': 'KnowledgeEngine', 'metadata': {'name': 'retriever1'}}}}],
         },
     ]
 

uv.lock

@@ -964,6 +964,30 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/bc/58/6b3d24e6b9bc474a2dcdee65dfd1f008867015408a271562e4b690561a4d/cryptography-46.0.5-pp311-pypy311_pp73-win_amd64.whl", hash = "sha256:8456928655f856c6e1533ff59d5be76578a7157224dbd9ce6872f25055ab9ab7", size = 3407605, upload-time = "2026-02-10T19:18:29.233Z" },
 ]
 
+[[package]]
+name = "cuda-bindings"
+version = "12.9.4"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "cuda-pathfinder", marker = "python_full_version >= '3.14' and sys_platform != 'emscripten' and sys_platform != 'win32'" },
+]
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/45/e7/b47792cc2d01c7e1d37c32402182524774dadd2d26339bd224e0e913832e/cuda_bindings-12.9.4-cp311-cp311-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:c912a3d9e6b6651853eed8eed96d6800d69c08e94052c292fec3f282c5a817c9", size = 12210593, upload-time = "2025-10-21T14:51:36.574Z" },
+    { url = "https://files.pythonhosted.org/packages/a9/c1/dabe88f52c3e3760d861401bb994df08f672ec893b8f7592dc91626adcf3/cuda_bindings-12.9.4-cp312-cp312-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:fda147a344e8eaeca0c6ff113d2851ffca8f7dfc0a6c932374ee5c47caa649c8", size = 12151019, upload-time = "2025-10-21T14:51:43.167Z" },
+    { url = "https://files.pythonhosted.org/packages/63/56/e465c31dc9111be3441a9ba7df1941fe98f4aa6e71e8788a3fb4534ce24d/cuda_bindings-12.9.4-cp313-cp313-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:32bdc5a76906be4c61eb98f546a6786c5773a881f3b166486449b5d141e4a39f", size = 11906628, upload-time = "2025-10-21T14:51:49.905Z" },
+    { url = "https://files.pythonhosted.org/packages/a3/84/1e6be415e37478070aeeee5884c2022713c1ecc735e6d82d744de0252eee/cuda_bindings-12.9.4-cp313-cp313t-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:56e0043c457a99ac473ddc926fe0dc4046694d99caef633e92601ab52cbe17eb", size = 11925991, upload-time = "2025-10-21T14:51:56.535Z" },
+    { url = "https://files.pythonhosted.org/packages/d1/af/6dfd8f2ed90b1d4719bc053ff8940e494640fe4212dc3dd72f383e4992da/cuda_bindings-12.9.4-cp314-cp314-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:8b72ee72a9cc1b531db31eebaaee5c69a8ec3500e32c6933f2d3b15297b53686", size = 11922703, upload-time = "2025-10-21T14:52:03.585Z" },
+    { url = "https://files.pythonhosted.org/packages/6c/19/90ac264acc00f6df8a49378eedec9fd2db3061bf9263bf9f39fd3d8377c3/cuda_bindings-12.9.4-cp314-cp314t-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:d80bffc357df9988dca279734bc9674c3934a654cab10cadeed27ce17d8635ee", size = 11924658, upload-time = "2025-10-21T14:52:10.411Z" },
+]
+
+[[package]]
+name = "cuda-pathfinder"
+version = "1.4.1"
+source = { registry = "https://pypi.org/simple" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/07/02/59a5bc738a09def0b49aea0e460bdf97f65206d0d041246147cf6207e69c/cuda_pathfinder-1.4.1-py3-none-any.whl", hash = "sha256:40793006082de88e0950753655e55558a446bed9a7d9d0bcb48b2506d50ed82a", size = 43903, upload-time = "2026-03-06T21:05:24.372Z" },
+]
+
 [[package]]
 name = "dashscope"
 version = "1.25.10"
@@ -1088,7 +1112,7 @@ wheels = [
 
 [[package]]
 name = "flask"
-version = "3.1.2"
+version = "3.1.3"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
     { name = "blinker" },
@@ -1098,9 +1122,9 @@ dependencies = [
     { name = "markupsafe" },
     { name = "werkzeug" },
 ]
-sdist = { url = "https://files.pythonhosted.org/packages/dc/6d/cfe3c0fcc5e477df242b98bfe186a4c34357b4847e87ecaef04507332dab/flask-3.1.2.tar.gz", hash = "sha256:bf656c15c80190ed628ad08cdfd3aaa35beb087855e2f494910aa3774cc4fd87", size = 720160, upload-time = "2025-08-19T21:03:21.205Z" }
+sdist = { url = "https://files.pythonhosted.org/packages/26/00/35d85dcce6c57fdc871f3867d465d780f302a175ea360f62533f12b27e2b/flask-3.1.3.tar.gz", hash = "sha256:0ef0e52b8a9cd932855379197dd8f94047b359ca0a78695144304cb45f87c9eb", size = 759004, upload-time = "2026-02-19T05:00:57.678Z" }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/ec/f9/7f9263c5695f4bd0023734af91bedb2ff8209e8de6ead162f35d8dc762fd/flask-3.1.2-py3-none-any.whl", hash = "sha256:ca1d8112ec8a6158cc29ea4858963350011b5c846a414cdb7a954aa9e967d03c", size = 103308, upload-time = "2025-08-19T21:03:19.499Z" },
+    { url = "https://files.pythonhosted.org/packages/7f/9c/34f6962f9b9e9c71f6e5ed806e0d0ff03c9d1b0b2340088a0cf4bce09b18/flask-3.1.3-py3-none-any.whl", hash = "sha256:f4bcbefc124291925f1a26446da31a5178f9483862233b23c0c96a20701f670c", size = 103424, upload-time = "2026-02-19T05:00:56.027Z" },
 ]
 
 [[package]]
@@ -1729,6 +1753,15 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/14/2f/967ba146e6d58cf6a652da73885f52fc68001525b4197effc174321d70b4/jmespath-1.1.0-py3-none-any.whl", hash = "sha256:a5663118de4908c91729bea0acadca56526eb2698e83de10cd116ae0f4e97c64", size = 20419, upload-time = "2026-01-22T16:35:24.919Z" },
 ]
 
+[[package]]
+name = "joblib"
+version = "1.5.3"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/41/f2/d34e8b3a08a9cc79a50b2208a93dce981fe615b64d5a4d4abee421d898df/joblib-1.5.3.tar.gz", hash = "sha256:8561a3269e6801106863fd0d6d84bb737be9e7631e33aaed3fb9ce5953688da3", size = 331603, upload-time = "2025-12-15T08:41:46.427Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/7b/91/984aca2ec129e2757d1e4e3c81c3fcda9d0f85b74670a094cc443d9ee949/joblib-1.5.3-py3-none-any.whl", hash = "sha256:5fc3c5039fc5ca8c0276333a188bbd59d6b7ab37fe6632daa76bc7f9ec18e713", size = 309071, upload-time = "2025-12-15T08:41:44.973Z" },
+]
+
 [[package]]
 name = "jsonpatch"
 version = "1.33"
@@ -1799,7 +1832,7 @@ wheels = [
 
 [[package]]
 name = "langbot"
-version = "4.8.6"
+version = "4.9.3"
 source = { editable = "." }
 dependencies = [
     { name = "aiocqhttp" },
@@ -1813,6 +1846,7 @@ dependencies = [
     { name = "asyncpg" },
     { name = "beautifulsoup4" },
     { name = "boto3" },
+    { name = "botocore" },
     { name = "certifi" },
     { name = "chardet" },
     { name = "chromadb" },
@@ -1891,9 +1925,10 @@ requires-dist = [
     { name = "asyncpg", specifier = ">=0.30.0" },
     { name = "beautifulsoup4", specifier = ">=4.12.3" },
     { name = "boto3", specifier = ">=1.35.0" },
+    { name = "botocore", specifier = ">=1.42.39" },
     { name = "certifi", specifier = ">=2025.4.26" },
     { name = "chardet", specifier = ">=5.2.0" },
-    { name = "chromadb", specifier = ">=0.4.24" },
+    { name = "chromadb", specifier = ">=1.0.0,<2.0.0" },
     { name = "colorlog", specifier = "~=6.6.0" },
     { name = "cryptography", specifier = ">=44.0.3" },
     { name = "dashscope", specifier = ">=1.25.10" },
@@ -1902,7 +1937,7 @@ requires-dist = [
     { name = "ebooklib", specifier = ">=0.18" },
     { name = "gewechat-client", specifier = ">=0.1.5" },
     { name = "html2text", specifier = ">=2024.2.26" },
-    { name = "langbot-plugin", specifier = "==0.2.7" },
+    { name = "langbot-plugin", specifier = "==0.3.2" },
     { name = "langchain", specifier = ">=0.2.0" },
     { name = "langchain-text-splitters", specifier = ">=0.0.1" },
     { name = "lark-oapi", specifier = ">=1.4.15" },
@@ -1925,7 +1960,7 @@ requires-dist = [
     { name = "pymilvus", specifier = ">=2.6.4" },
     { name = "pynacl", specifier = ">=1.5.0" },
     { name = "pypdf2", specifier = ">=3.0.1" },
-    { name = "pyseekdb", specifier = "==1.0.0b7" },
+    { name = "pyseekdb", specifier = "==1.1.0.post3" },
     { name = "python-docx", specifier = ">=1.1.0" },
     { name = "python-socks", specifier = ">=2.7.1" },
     { name = "python-telegram-bot", specifier = ">=22.0" },
@@ -1958,7 +1993,7 @@ dev = [
 
 [[package]]
 name = "langbot-plugin"
-version = "0.2.7"
+version = "0.3.2"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
     { name = "aiofiles" },
@@ -1976,28 +2011,28 @@ dependencies = [
     { name = "watchdog" },
     { name = "websockets" },
 ]
-sdist = { url = "https://files.pythonhosted.org/packages/9e/a0/babd76596e5de38149da67b8da20e0519cc5f10080de9dc2b16919486f29/langbot_plugin-0.2.7.tar.gz", hash = "sha256:5c8ad1820283901a33356f79a56c84b4744712a463e1c7aecc6e9defe4db4446", size = 162458, upload-time = "2026-02-25T06:00:52.512Z" }
+sdist = { url = "https://files.pythonhosted.org/packages/05/32/1b029961d718b5418d9d139687287193d4657914c467833176d9c060fb4c/langbot_plugin-0.3.2.tar.gz", hash = "sha256:2f7f16285600ec019a4e8cc8b40bc8f8d404e3bbc69c9d129620dc70b3bde2f8", size = 170431, upload-time = "2026-03-14T12:42:55.175Z" }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/32/2a/6575cf5d5babb7a9400a8aca243e4b8341d83b673e5e9c0394c0393f1c3e/langbot_plugin-0.2.7-py3-none-any.whl", hash = "sha256:17344e61537a5bb97fc77cd83812b5db926f29005e92fefbcbaca5bb47bf55f0", size = 133476, upload-time = "2026-02-25T06:00:50.988Z" },
+    { url = "https://files.pythonhosted.org/packages/91/d7/a5dd6cff000a4f7fe88692689be276875b04485f48fc26e162f1ad2a341c/langbot_plugin-0.3.2-py3-none-any.whl", hash = "sha256:a1de527c4d651e6b6ba2458b6fac09a300e6b1ffdc5a25bbfad88d970cfd6cfd", size = 144906, upload-time = "2026-03-14T12:42:56.486Z" },
 ]
 
 [[package]]
 name = "langchain"
-version = "1.2.7"
+version = "1.2.12"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
     { name = "langchain-core" },
     { name = "langgraph" },
     { name = "pydantic" },
 ]
-sdist = { url = "https://files.pythonhosted.org/packages/47/f2/478ca9f3455b5d66402066d287eae7e8d6c722acfb8553937e06af708334/langchain-1.2.7.tar.gz", hash = "sha256:ba40e8d5b069a22f7085f54f405973da3d87cfdebf116282e77c692271432ecb", size = 556837, upload-time = "2026-01-23T15:22:10.817Z" }
+sdist = { url = "https://files.pythonhosted.org/packages/d8/1d/1af2fc0ac084d4781778b7846b1aed62e05006bf2d73fdf84ac3a8f5225c/langchain-1.2.12.tar.gz", hash = "sha256:ed705b5b293799f7e3e394387f398a1b71707542758283206c8c21415759d991", size = 566444, upload-time = "2026-03-11T22:21:00.712Z" }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/dd/c8/9ce37ae34870834c7d00bb14ff4876b700db31b928635e3307804dc41d74/langchain-1.2.7-py3-none-any.whl", hash = "sha256:1d643c8ca569bcde2470b853807f74f0768b3982d25d66d57db21a166aabda72", size = 108827, upload-time = "2026-01-23T15:22:09.771Z" },
+    { url = "https://files.pythonhosted.org/packages/ca/51/09bb1cfb0b57ae9440ca56cc576e4dc792f83d030eef7637d2c516dcb0a0/langchain-1.2.12-py3-none-any.whl", hash = "sha256:60eff184b8f92c2610f5a4c9a97ad339a891adb01901e83e4df8e6c9c69cf852", size = 112373, upload-time = "2026-03-11T22:20:59.508Z" },
 ]
 
 [[package]]
 name = "langchain-core"
-version = "1.2.7"
+version = "1.2.18"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
     { name = "jsonpatch" },
@@ -2009,9 +2044,9 @@ dependencies = [
     { name = "typing-extensions" },
     { name = "uuid-utils" },
 ]
-sdist = { url = "https://files.pythonhosted.org/packages/a2/0e/664d8d81b3493e09cbab72448d2f9d693d1fa5aa2bcc488602203a9b6da0/langchain_core-1.2.7.tar.gz", hash = "sha256:e1460639f96c352b4a41c375f25aeb8d16ffc1769499fb1c20503aad59305ced", size = 837039, upload-time = "2026-01-09T17:44:25.505Z" }
+sdist = { url = "https://files.pythonhosted.org/packages/18/b7/8bbd0d99a6441b35d891e4b79e7d24c67722cdd363893ae650f24808cf5a/langchain_core-1.2.18.tar.gz", hash = "sha256:ffe53eec44636d092895b9fe25d28af3aaf79060e293fa7cda2a5aaa50c80d21", size = 836725, upload-time = "2026-03-09T20:40:07.229Z" }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/6e/6f/34a9fba14d191a67f7e2ee3dbce3e9b86d2fa7310e2c7f2c713583481bd2/langchain_core-1.2.7-py3-none-any.whl", hash = "sha256:452f4fef7a3d883357b22600788d37e3d8854ef29da345b7ac7099f33c31828b", size = 490232, upload-time = "2026-01-09T17:44:24.236Z" },
+    { url = "https://files.pythonhosted.org/packages/1f/d8/9418564ed4ab4f150668b25cf8c188266267d829362e9c9106946afa628b/langchain_core-1.2.18-py3-none-any.whl", hash = "sha256:cccb79523e0045174ab826054e555fddc973266770e427588c8f1ec9d9d6212b", size = 503048, upload-time = "2026-03-09T20:40:06.115Z" },
 ]
 
 [[package]]
@@ -2028,7 +2063,7 @@ wheels = [
 
 [[package]]
 name = "langgraph"
-version = "1.0.7"
+version = "1.1.1"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
     { name = "langchain-core" },
@@ -2038,9 +2073,9 @@ dependencies = [
     { name = "pydantic" },
     { name = "xxhash" },
 ]
-sdist = { url = "https://files.pythonhosted.org/packages/72/5b/f72655717c04e33d3b62f21b166dc063d192b53980e9e3be0e2a117f1c9f/langgraph-1.0.7.tar.gz", hash = "sha256:0cfdfee51e6e8cfe503ecc7367c73933437c505b03fa10a85c710975c8182d9a", size = 497098, upload-time = "2026-01-22T16:57:47.303Z" }
+sdist = { url = "https://files.pythonhosted.org/packages/6d/1a/6dbad0c87fb39a58e5ced85297511cc4bcad06cc420b20898eecafece2a2/langgraph-1.1.1.tar.gz", hash = "sha256:cd6282efc657c955b41bff6bd9693de58137ad18f7e7f16b4d17c7d2118d53e1", size = 544040, upload-time = "2026-03-11T22:14:47.845Z" }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/7e/0e/fe80144e3e4048e5d19ccdb91ac547c1a7dc3da8dbd1443e210048194c14/langgraph-1.0.7-py3-none-any.whl", hash = "sha256:9d68e8f8dd8f3de2fec45f9a06de05766d9b075b78fb03171779893b7a52c4d2", size = 157353, upload-time = "2026-01-22T16:57:45.997Z" },
+    { url = "https://files.pythonhosted.org/packages/dc/c1/572187bb61a534050ef2d5030e7abe46b19694ec106604fe12ddcb8672c7/langgraph-1.1.1-py3-none-any.whl", hash = "sha256:d0cc8d347131cbfc010e65aad9b0f1afbd0e151f470c288bec1f3df8336c50c6", size = 167502, upload-time = "2026-03-11T22:14:46.121Z" },
 ]
 
 [[package]]
@@ -2058,15 +2093,15 @@ wheels = [
 
 [[package]]
 name = "langgraph-prebuilt"
-version = "1.0.7"
+version = "1.0.8"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
     { name = "langchain-core" },
     { name = "langgraph-checkpoint" },
 ]
-sdist = { url = "https://files.pythonhosted.org/packages/a7/59/711aecd1a50999456850dc328f3cad72b4372d8218838d8d5326f80cb76f/langgraph_prebuilt-1.0.7.tar.gz", hash = "sha256:38e097e06de810de4d0e028ffc0e432bb56d1fb417620fb1dfdc76c5e03e4bf9", size = 163692, upload-time = "2026-01-22T16:45:22.801Z" }
+sdist = { url = "https://files.pythonhosted.org/packages/0d/06/dd61a5c2dce009d1b03b1d56f2a85b3127659fdddf5b3be5d8f1d60820fb/langgraph_prebuilt-1.0.8.tar.gz", hash = "sha256:0cd3cf5473ced8a6cd687cc5294e08d3de57529d8dd14fdc6ae4899549efcf69", size = 164442, upload-time = "2026-02-19T18:14:39.083Z" }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/47/49/5e37abb3f38a17a3487634abc2a5da87c208cc1d14577eb8d7184b25c886/langgraph_prebuilt-1.0.7-py3-none-any.whl", hash = "sha256:e14923516504405bb5edc3977085bc9622c35476b50c1808544490e13871fe7c", size = 35324, upload-time = "2026-01-22T16:45:21.784Z" },
+    { url = "https://files.pythonhosted.org/packages/dc/41/ec966424ad3f2ed3996d24079d3342c8cd6c0bd0653c12b2a917a685ec6c/langgraph_prebuilt-1.0.8-py3-none-any.whl", hash = "sha256:d16a731e591ba4470f3e313a319c7eee7dbc40895bcf15c821f985a3522a7ce0", size = 35648, upload-time = "2026-02-19T18:14:37.611Z" },
 ]
 
 [[package]]
@@ -2814,6 +2849,15 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/4a/67/5c9c8f1ba4a599e35a77ca7e0a0210ab6cd732f719bc3b0fc95c69aaca10/nakuru_project_idk-0.0.2.1-py3-none-any.whl", hash = "sha256:bddd8af8a46ef381bd05b806d6c07bd8ba407c58b47ce6148d750bd77c4420bc", size = 24281, upload-time = "2023-05-07T15:00:25.094Z" },
 ]
 
+[[package]]
+name = "networkx"
+version = "3.6.1"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/6a/51/63fe664f3908c97be9d2e4f1158eb633317598cfa6e1fc14af5383f17512/networkx-3.6.1.tar.gz", hash = "sha256:26b7c357accc0c8cde558ad486283728b65b6a95d85ee1cd66bafab4c8168509", size = 2517025, upload-time = "2025-12-08T17:02:39.908Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/9e/c9/b2622292ea83fbb4ec318f5b9ab867d0a28ab43c5717bb85b0a5f6b3b0a4/networkx-3.6.1-py3-none-any.whl", hash = "sha256:d47fbf302e7d9cbbb9e2555a0d267983d2aa476bac30e90dfbe5669bd57f3762", size = 2068504, upload-time = "2025-12-08T17:02:38.159Z" },
+]
+
 [[package]]
 name = "nodeenv"
 version = "1.10.0"
@@ -2902,6 +2946,140 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/de/e5/b7d20451657664b07986c2f6e3be564433f5dcaf3482d68eaecd79afaf03/numpy-2.4.2-pp311-pypy311_pp73-win_amd64.whl", hash = "sha256:be71bf1edb48ebbbf7f6337b5bfd2f895d1902f6335a5830b20141fc126ffba0", size = 12502577, upload-time = "2026-01-31T23:13:07.08Z" },
 ]
 
+[[package]]
+name = "nvidia-cublas-cu12"
+version = "12.8.4.1"
+source = { registry = "https://pypi.org/simple" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/dc/61/e24b560ab2e2eaeb3c839129175fb330dfcfc29e5203196e5541a4c44682/nvidia_cublas_cu12-12.8.4.1-py3-none-manylinux_2_27_x86_64.whl", hash = "sha256:8ac4e771d5a348c551b2a426eda6193c19aa630236b418086020df5ba9667142", size = 594346921, upload-time = "2025-03-07T01:44:31.254Z" },
+]
+
+[[package]]
+name = "nvidia-cuda-cupti-cu12"
+version = "12.8.90"
+source = { registry = "https://pypi.org/simple" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/f8/02/2adcaa145158bf1a8295d83591d22e4103dbfd821bcaf6f3f53151ca4ffa/nvidia_cuda_cupti_cu12-12.8.90-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:ea0cb07ebda26bb9b29ba82cda34849e73c166c18162d3913575b0c9db9a6182", size = 10248621, upload-time = "2025-03-07T01:40:21.213Z" },
+]
+
+[[package]]
+name = "nvidia-cuda-nvrtc-cu12"
+version = "12.8.93"
+source = { registry = "https://pypi.org/simple" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/05/6b/32f747947df2da6994e999492ab306a903659555dddc0fbdeb9d71f75e52/nvidia_cuda_nvrtc_cu12-12.8.93-py3-none-manylinux2010_x86_64.manylinux_2_12_x86_64.whl", hash = "sha256:a7756528852ef889772a84c6cd89d41dfa74667e24cca16bb31f8f061e3e9994", size = 88040029, upload-time = "2025-03-07T01:42:13.562Z" },
+]
+
+[[package]]
+name = "nvidia-cuda-runtime-cu12"
+version = "12.8.90"
+source = { registry = "https://pypi.org/simple" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/0d/9b/a997b638fcd068ad6e4d53b8551a7d30fe8b404d6f1804abf1df69838932/nvidia_cuda_runtime_cu12-12.8.90-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:adade8dcbd0edf427b7204d480d6066d33902cab2a4707dcfc48a2d0fd44ab90", size = 954765, upload-time = "2025-03-07T01:40:01.615Z" },
+]
+
+[[package]]
+name = "nvidia-cudnn-cu12"
+version = "9.10.2.21"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "nvidia-cublas-cu12", marker = "python_full_version >= '3.14' and sys_platform != 'emscripten' and sys_platform != 'win32'" },
+]
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/ba/51/e123d997aa098c61d029f76663dedbfb9bc8dcf8c60cbd6adbe42f76d049/nvidia_cudnn_cu12-9.10.2.21-py3-none-manylinux_2_27_x86_64.whl", hash = "sha256:949452be657fa16687d0930933f032835951ef0892b37d2d53824d1a84dc97a8", size = 706758467, upload-time = "2025-06-06T21:54:08.597Z" },
+]
+
+[[package]]
+name = "nvidia-cufft-cu12"
+version = "11.3.3.83"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "nvidia-nvjitlink-cu12", marker = "python_full_version >= '3.14' and sys_platform != 'emscripten' and sys_platform != 'win32'" },
+]
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/1f/13/ee4e00f30e676b66ae65b4f08cb5bcbb8392c03f54f2d5413ea99a5d1c80/nvidia_cufft_cu12-11.3.3.83-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:4d2dd21ec0b88cf61b62e6b43564355e5222e4a3fb394cac0db101f2dd0d4f74", size = 193118695, upload-time = "2025-03-07T01:45:27.821Z" },
+]
+
+[[package]]
+name = "nvidia-cufile-cu12"
+version = "1.13.1.3"
+source = { registry = "https://pypi.org/simple" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/bb/fe/1bcba1dfbfb8d01be8d93f07bfc502c93fa23afa6fd5ab3fc7c1df71038a/nvidia_cufile_cu12-1.13.1.3-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:1d069003be650e131b21c932ec3d8969c1715379251f8d23a1860554b1cb24fc", size = 1197834, upload-time = "2025-03-07T01:45:50.723Z" },
+]
+
+[[package]]
+name = "nvidia-curand-cu12"
+version = "10.3.9.90"
+source = { registry = "https://pypi.org/simple" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/fb/aa/6584b56dc84ebe9cf93226a5cde4d99080c8e90ab40f0c27bda7a0f29aa1/nvidia_curand_cu12-10.3.9.90-py3-none-manylinux_2_27_x86_64.whl", hash = "sha256:b32331d4f4df5d6eefa0554c565b626c7216f87a06a4f56fab27c3b68a830ec9", size = 63619976, upload-time = "2025-03-07T01:46:23.323Z" },
+]
+
+[[package]]
+name = "nvidia-cusolver-cu12"
+version = "11.7.3.90"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "nvidia-cublas-cu12", marker = "python_full_version >= '3.14' and sys_platform != 'emscripten' and sys_platform != 'win32'" },
+    { name = "nvidia-cusparse-cu12", marker = "python_full_version >= '3.14' and sys_platform != 'emscripten' and sys_platform != 'win32'" },
+    { name = "nvidia-nvjitlink-cu12", marker = "python_full_version >= '3.14' and sys_platform != 'emscripten' and sys_platform != 'win32'" },
+]
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/85/48/9a13d2975803e8cf2777d5ed57b87a0b6ca2cc795f9a4f59796a910bfb80/nvidia_cusolver_cu12-11.7.3.90-py3-none-manylinux_2_27_x86_64.whl", hash = "sha256:4376c11ad263152bd50ea295c05370360776f8c3427b30991df774f9fb26c450", size = 267506905, upload-time = "2025-03-07T01:47:16.273Z" },
+]
+
+[[package]]
+name = "nvidia-cusparse-cu12"
+version = "12.5.8.93"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "nvidia-nvjitlink-cu12", marker = "python_full_version >= '3.14' and sys_platform != 'emscripten' and sys_platform != 'win32'" },
+]
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/c2/f5/e1854cb2f2bcd4280c44736c93550cc300ff4b8c95ebe370d0aa7d2b473d/nvidia_cusparse_cu12-12.5.8.93-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:1ec05d76bbbd8b61b06a80e1eaf8cf4959c3d4ce8e711b65ebd0443bb0ebb13b", size = 288216466, upload-time = "2025-03-07T01:48:13.779Z" },
+]
+
+[[package]]
+name = "nvidia-cusparselt-cu12"
+version = "0.7.1"
+source = { registry = "https://pypi.org/simple" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/56/79/12978b96bd44274fe38b5dde5cfb660b1d114f70a65ef962bcbbed99b549/nvidia_cusparselt_cu12-0.7.1-py3-none-manylinux2014_x86_64.whl", hash = "sha256:f1bb701d6b930d5a7cea44c19ceb973311500847f81b634d802b7b539dc55623", size = 287193691, upload-time = "2025-02-26T00:15:44.104Z" },
+]
+
+[[package]]
+name = "nvidia-nccl-cu12"
+version = "2.27.5"
+source = { registry = "https://pypi.org/simple" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/6e/89/f7a07dc961b60645dbbf42e80f2bc85ade7feb9a491b11a1e973aa00071f/nvidia_nccl_cu12-2.27.5-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:ad730cf15cb5d25fe849c6e6ca9eb5b76db16a80f13f425ac68d8e2e55624457", size = 322348229, upload-time = "2025-06-26T04:11:28.385Z" },
+]
+
+[[package]]
+name = "nvidia-nvjitlink-cu12"
+version = "12.8.93"
+source = { registry = "https://pypi.org/simple" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/f6/74/86a07f1d0f42998ca31312f998bd3b9a7eff7f52378f4f270c8679c77fb9/nvidia_nvjitlink_cu12-12.8.93-py3-none-manylinux2010_x86_64.manylinux_2_12_x86_64.whl", hash = "sha256:81ff63371a7ebd6e6451970684f916be2eab07321b73c9d244dc2b4da7f73b88", size = 39254836, upload-time = "2025-03-07T01:49:55.661Z" },
+]
+
+[[package]]
+name = "nvidia-nvshmem-cu12"
+version = "3.4.5"
+source = { registry = "https://pypi.org/simple" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/b5/09/6ea3ea725f82e1e76684f0708bbedd871fc96da89945adeba65c3835a64c/nvidia_nvshmem_cu12-3.4.5-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:042f2500f24c021db8a06c5eec2539027d57460e1c1a762055a6554f72c369bd", size = 139103095, upload-time = "2025-09-06T00:32:31.266Z" },
+]
+
+[[package]]
+name = "nvidia-nvtx-cu12"
+version = "12.8.90"
+source = { registry = "https://pypi.org/simple" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/a2/eb/86626c1bbc2edb86323022371c39aa48df6fd8b0a1647bc274577f72e90b/nvidia_nvtx_cu12-12.8.90-py3-none-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:5b17e2001cc0d751a5bc2c6ec6d26ad95913324a4adb86788c944f8ce9ba441f", size = 89954, upload-time = "2025-03-07T01:42:44.131Z" },
+]
+
 [[package]]
 name = "oauthlib"
 version = "3.3.1"
@@ -3922,12 +4100,16 @@ name = "pylibseekdb"
 version = "1.1.0"
 source = { registry = "https://pypi.org/simple" }
 wheels = [
+    { url = "https://files.pythonhosted.org/packages/1c/b8/c226744a7a1da9295725920a36867ee5665f2617972c7881d5ed4cbd45c8/pylibseekdb-1.1.0-cp311-cp311-macosx_15_0_arm64.whl", hash = "sha256:0a0ad03d87f1db1a7087ba89e398ce1ee00496e977d38c493104d0d517590968", size = 148743770, upload-time = "2026-01-30T05:26:14.275Z" },
     { url = "https://files.pythonhosted.org/packages/51/4d/57151735afc29039f4ed680256012a33dd719ba3fd84d7c33a9bd260fc8a/pylibseekdb-1.1.0-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:e272bee013aabab152c4795676b3b0ba1107a8058f29a07d2a803168faea090c", size = 147132528, upload-time = "2026-01-30T03:40:10.878Z" },
     { url = "https://files.pythonhosted.org/packages/88/d7/5583fbf27e89952cda52bb9b1919229bd652d02aafac156758ac862c48e7/pylibseekdb-1.1.0-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:116a28356532705ed262e2a7951ac8221ae8c97ade866fdab2df521dcca62530", size = 170696822, upload-time = "2026-01-30T03:40:18.417Z" },
+    { url = "https://files.pythonhosted.org/packages/5d/2b/150592287119f80cff9b025d59879a561a0cca80e71cecbf74a41af6220b/pylibseekdb-1.1.0-cp312-cp312-macosx_15_0_arm64.whl", hash = "sha256:d6ae33353e833cb56a7ce2cdb0305b872cdac9467eb79c277f82479c529b38ef", size = 148734111, upload-time = "2026-01-30T05:26:56.906Z" },
     { url = "https://files.pythonhosted.org/packages/b8/a3/b55087293115ecbe22313b40533fd67b0192c36e6bedb05aa7058a83a86a/pylibseekdb-1.1.0-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:9e2f8240b08a93e347d32534e7c394b7a151b67555a384eb88d73d4b0f8b9d14", size = 147137592, upload-time = "2026-01-30T03:40:26.087Z" },
     { url = "https://files.pythonhosted.org/packages/04/31/c0979960d790621dec277f64b5d6c70932f8bb9adb59029d7b481cfe9c30/pylibseekdb-1.1.0-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:4d8615471bac39b1980951cbce0d742fa7bec676f28eb95f4db687fdd1e9c71b", size = 170681044, upload-time = "2026-01-30T03:40:34.276Z" },
+    { url = "https://files.pythonhosted.org/packages/33/7d/8acbf3eca93905c1b13b015a9e02b426fc69c10e7c162be96b35a2b1c7a4/pylibseekdb-1.1.0-cp313-cp313-macosx_15_0_arm64.whl", hash = "sha256:d5688a0fe6fc703e5a707cbe0e139d570f1d34daff1491304d6b43154f2e12d9", size = 148743750, upload-time = "2026-01-30T05:27:39.832Z" },
     { url = "https://files.pythonhosted.org/packages/c8/24/7f510ad13ad129a691fa965dc5bce874320b682674cbf12fc2e35310719b/pylibseekdb-1.1.0-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:1e53d171246239bd526d1a1f9b3abef1ad9b10597bc1c0a2acf7e65afbd7d844", size = 147136041, upload-time = "2026-01-30T03:40:41.782Z" },
     { url = "https://files.pythonhosted.org/packages/ed/eb/c5988e1ad72233a920f4e444d8d866c42363220b340d78a7525307922f35/pylibseekdb-1.1.0-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:66d01ee9c0ad4a2e88ea2420f9c4d1ee9bb011b70c553a654c8a4e230e920ad7", size = 170684140, upload-time = "2026-01-30T03:40:49.351Z" },
+    { url = "https://files.pythonhosted.org/packages/9a/6f/b4a619c3a1b937fb080aa977b1d4011a1e587255707d54856188e5359a4c/pylibseekdb-1.1.0-cp314-cp314-macosx_15_0_arm64.whl", hash = "sha256:11d2fbc98dcb8ec97257b949184dc09d9ba693811e77457bba9c8f80d282c265", size = 148745880, upload-time = "2026-01-30T05:38:26.631Z" },
     { url = "https://files.pythonhosted.org/packages/0c/94/534359608571d08825ac21e709aa680b559989c905f99e273d82d5b17db2/pylibseekdb-1.1.0-cp314-cp314-manylinux_2_28_aarch64.whl", hash = "sha256:ff05ac4bb13a4b5f9dd03771ded866beed72562ea497f68a4ae897c226afc446", size = 147132460, upload-time = "2026-01-30T03:40:56.684Z" },
     { url = "https://files.pythonhosted.org/packages/19/5e/7588a06918ac145fb69e57ae372b72d6fc713b9263c29eb7268f8a4edbef/pylibseekdb-1.1.0-cp314-cp314-manylinux_2_28_x86_64.whl", hash = "sha256:065158b79192cce7635995a7599e99b21a3ff729cd6f68e31a65ed62f830bd3a", size = 170677921, upload-time = "2026-01-30T03:41:03.783Z" },
 ]
@@ -4041,20 +4223,21 @@ wheels = [
 
 [[package]]
 name = "pyseekdb"
-version = "1.0.0b7"
+version = "1.1.0.post3"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
-    { name = "httpx" },
+    { name = "httpx", marker = "python_full_version < '3.14'" },
     { name = "numpy" },
-    { name = "onnxruntime" },
-    { name = "pylibseekdb", marker = "sys_platform == 'linux'" },
+    { name = "onnxruntime", marker = "python_full_version < '3.14'" },
+    { name = "pylibseekdb", marker = "(platform_machine == 'arm64' and sys_platform == 'darwin') or sys_platform == 'linux'" },
     { name = "pymysql" },
+    { name = "sentence-transformers", marker = "python_full_version >= '3.14'" },
     { name = "tenacity" },
-    { name = "tokenizers" },
-    { name = "tqdm" },
+    { name = "tokenizers", marker = "python_full_version < '3.14'" },
+    { name = "tqdm", marker = "python_full_version < '3.14'" },
 ]
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/92/6a/a0d4728de90e028a60a3583e6e96579087f0cf793e705ea7898a1490541c/pyseekdb-1.0.0b7-py3-none-any.whl", hash = "sha256:e32920636c345bc73adf03040f9bcb1ecc420d652cedae1558999cce19a67d52", size = 60927, upload-time = "2025-12-29T13:19:04.669Z" },
+    { url = "https://files.pythonhosted.org/packages/58/6e/2373239ab80c35a17aa14e8219727f06567e91d3b7f1b8c36d28ce94d04b/pyseekdb-1.1.0.post3-py3-none-any.whl", hash = "sha256:0437c9a4de72be44eb24b070b2b8099086467c08af10a57191498a61257a4bfb", size = 110985, upload-time = "2026-02-12T14:19:05.402Z" },
 ]
 
 [[package]]
@@ -4634,6 +4817,168 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/fc/51/727abb13f44c1fcf6d145979e1535a35794db0f6e450a0cb46aa24732fe2/s3transfer-0.16.0-py3-none-any.whl", hash = "sha256:18e25d66fed509e3868dc1572b3f427ff947dd2c56f844a5bf09481ad3f3b2fe", size = 86830, upload-time = "2025-12-01T02:30:57.729Z" },
 ]
 
+[[package]]
+name = "safetensors"
+version = "0.7.0"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/29/9c/6e74567782559a63bd040a236edca26fd71bc7ba88de2ef35d75df3bca5e/safetensors-0.7.0.tar.gz", hash = "sha256:07663963b67e8bd9f0b8ad15bb9163606cd27cc5a1b96235a50d8369803b96b0", size = 200878, upload-time = "2025-11-19T15:18:43.199Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/fa/47/aef6c06649039accf914afef490268e1067ed82be62bcfa5b7e886ad15e8/safetensors-0.7.0-cp38-abi3-macosx_10_12_x86_64.whl", hash = "sha256:c82f4d474cf725255d9e6acf17252991c3c8aac038d6ef363a4bf8be2f6db517", size = 467781, upload-time = "2025-11-19T15:18:35.84Z" },
+    { url = "https://files.pythonhosted.org/packages/e8/00/374c0c068e30cd31f1e1b46b4b5738168ec79e7689ca82ee93ddfea05109/safetensors-0.7.0-cp38-abi3-macosx_11_0_arm64.whl", hash = "sha256:94fd4858284736bb67a897a41608b5b0c2496c9bdb3bf2af1fa3409127f20d57", size = 447058, upload-time = "2025-11-19T15:18:34.416Z" },
+    { url = "https://files.pythonhosted.org/packages/f1/06/578ffed52c2296f93d7fd2d844cabfa92be51a587c38c8afbb8ae449ca89/safetensors-0.7.0-cp38-abi3-manylinux_2_17_aarch64.manylinux2014_aarch64.whl", hash = "sha256:e07d91d0c92a31200f25351f4acb2bc6aff7f48094e13ebb1d0fb995b54b6542", size = 491748, upload-time = "2025-11-19T15:18:09.79Z" },
+    { url = "https://files.pythonhosted.org/packages/ae/33/1debbbb70e4791dde185edb9413d1fe01619255abb64b300157d7f15dddd/safetensors-0.7.0-cp38-abi3-manylinux_2_17_armv7l.manylinux2014_armv7l.whl", hash = "sha256:8469155f4cb518bafb4acf4865e8bb9d6804110d2d9bdcaa78564b9fd841e104", size = 503881, upload-time = "2025-11-19T15:18:16.145Z" },
+    { url = "https://files.pythonhosted.org/packages/8e/1c/40c2ca924d60792c3be509833df711b553c60effbd91da6f5284a83f7122/safetensors-0.7.0-cp38-abi3-manylinux_2_17_ppc64le.manylinux2014_ppc64le.whl", hash = "sha256:54bef08bf00a2bff599982f6b08e8770e09cc012d7bba00783fc7ea38f1fb37d", size = 623463, upload-time = "2025-11-19T15:18:21.11Z" },
+    { url = "https://files.pythonhosted.org/packages/9b/3a/13784a9364bd43b0d61eef4bea2845039bc2030458b16594a1bd787ae26e/safetensors-0.7.0-cp38-abi3-manylinux_2_17_s390x.manylinux2014_s390x.whl", hash = "sha256:42cb091236206bb2016d245c377ed383aa7f78691748f3bb6ee1bfa51ae2ce6a", size = 532855, upload-time = "2025-11-19T15:18:25.719Z" },
+    { url = "https://files.pythonhosted.org/packages/a0/60/429e9b1cb3fc651937727befe258ea24122d9663e4d5709a48c9cbfceecb/safetensors-0.7.0-cp38-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:dac7252938f0696ddea46f5e855dd3138444e82236e3be475f54929f0c510d48", size = 507152, upload-time = "2025-11-19T15:18:33.023Z" },
+    { url = "https://files.pythonhosted.org/packages/3c/a8/4b45e4e059270d17af60359713ffd83f97900d45a6afa73aaa0d737d48b6/safetensors-0.7.0-cp38-abi3-manylinux_2_5_i686.manylinux1_i686.whl", hash = "sha256:1d060c70284127fa805085d8f10fbd0962792aed71879d00864acda69dbab981", size = 541856, upload-time = "2025-11-19T15:18:31.075Z" },
+    { url = "https://files.pythonhosted.org/packages/06/87/d26d8407c44175d8ae164a95b5a62707fcc445f3c0c56108e37d98070a3d/safetensors-0.7.0-cp38-abi3-musllinux_1_2_aarch64.whl", hash = "sha256:cdab83a366799fa730f90a4ebb563e494f28e9e92c4819e556152ad55e43591b", size = 674060, upload-time = "2025-11-19T15:18:37.211Z" },
+    { url = "https://files.pythonhosted.org/packages/11/f5/57644a2ff08dc6325816ba7217e5095f17269dada2554b658442c66aed51/safetensors-0.7.0-cp38-abi3-musllinux_1_2_armv7l.whl", hash = "sha256:672132907fcad9f2aedcb705b2d7b3b93354a2aec1b2f706c4db852abe338f85", size = 771715, upload-time = "2025-11-19T15:18:38.689Z" },
+    { url = "https://files.pythonhosted.org/packages/86/31/17883e13a814bd278ae6e266b13282a01049b0c81341da7fd0e3e71a80a3/safetensors-0.7.0-cp38-abi3-musllinux_1_2_i686.whl", hash = "sha256:5d72abdb8a4d56d4020713724ba81dac065fedb7f3667151c4a637f1d3fb26c0", size = 714377, upload-time = "2025-11-19T15:18:40.162Z" },
+    { url = "https://files.pythonhosted.org/packages/4a/d8/0c8a7dc9b41dcac53c4cbf9df2b9c83e0e0097203de8b37a712b345c0be5/safetensors-0.7.0-cp38-abi3-musllinux_1_2_x86_64.whl", hash = "sha256:b0f6d66c1c538d5a94a73aa9ddca8ccc4227e6c9ff555322ea40bdd142391dd4", size = 677368, upload-time = "2025-11-19T15:18:41.627Z" },
+    { url = "https://files.pythonhosted.org/packages/05/e5/cb4b713c8a93469e3c5be7c3f8d77d307e65fe89673e731f5c2bfd0a9237/safetensors-0.7.0-cp38-abi3-win32.whl", hash = "sha256:c74af94bf3ac15ac4d0f2a7c7b4663a15f8c2ab15ed0fc7531ca61d0835eccba", size = 326423, upload-time = "2025-11-19T15:18:45.74Z" },
+    { url = "https://files.pythonhosted.org/packages/5d/e6/ec8471c8072382cb91233ba7267fd931219753bb43814cbc71757bfd4dab/safetensors-0.7.0-cp38-abi3-win_amd64.whl", hash = "sha256:d1239932053f56f3456f32eb9625590cc7582e905021f94636202a864d470755", size = 341380, upload-time = "2025-11-19T15:18:44.427Z" },
+]
+
+[[package]]
+name = "scikit-learn"
+version = "1.8.0"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "joblib", marker = "python_full_version >= '3.14'" },
+    { name = "numpy", marker = "python_full_version >= '3.14'" },
+    { name = "scipy", marker = "python_full_version >= '3.14'" },
+    { name = "threadpoolctl", marker = "python_full_version >= '3.14'" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/0e/d4/40988bf3b8e34feec1d0e6a051446b1f66225f8529b9309becaeef62b6c4/scikit_learn-1.8.0.tar.gz", hash = "sha256:9bccbb3b40e3de10351f8f5068e105d0f4083b1a65fa07b6634fbc401a6287fd", size = 7335585, upload-time = "2025-12-10T07:08:53.618Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/c9/92/53ea2181da8ac6bf27170191028aee7251f8f841f8d3edbfdcaf2008fde9/scikit_learn-1.8.0-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:146b4d36f800c013d267b29168813f7a03a43ecd2895d04861f1240b564421da", size = 8595835, upload-time = "2025-12-10T07:07:39.385Z" },
+    { url = "https://files.pythonhosted.org/packages/01/18/d154dc1638803adf987910cdd07097d9c526663a55666a97c124d09fb96a/scikit_learn-1.8.0-cp311-cp311-macosx_12_0_arm64.whl", hash = "sha256:f984ca4b14914e6b4094c5d52a32ea16b49832c03bd17a110f004db3c223e8e1", size = 8080381, upload-time = "2025-12-10T07:07:41.93Z" },
+    { url = "https://files.pythonhosted.org/packages/8a/44/226142fcb7b7101e64fdee5f49dbe6288d4c7af8abf593237b70fca080a4/scikit_learn-1.8.0-cp311-cp311-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:5e30adb87f0cc81c7690a84f7932dd66be5bac57cfe16b91cb9151683a4a2d3b", size = 8799632, upload-time = "2025-12-10T07:07:43.899Z" },
+    { url = "https://files.pythonhosted.org/packages/36/4d/4a67f30778a45d542bbea5db2dbfa1e9e100bf9ba64aefe34215ba9f11f6/scikit_learn-1.8.0-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:ada8121bcb4dac28d930febc791a69f7cb1673c8495e5eee274190b73a4559c1", size = 9103788, upload-time = "2025-12-10T07:07:45.982Z" },
+    { url = "https://files.pythonhosted.org/packages/89/3c/45c352094cfa60050bcbb967b1faf246b22e93cb459f2f907b600f2ceda5/scikit_learn-1.8.0-cp311-cp311-win_amd64.whl", hash = "sha256:c57b1b610bd1f40ba43970e11ce62821c2e6569e4d74023db19c6b26f246cb3b", size = 8081706, upload-time = "2025-12-10T07:07:48.111Z" },
+    { url = "https://files.pythonhosted.org/packages/3d/46/5416595bb395757f754feb20c3d776553a386b661658fb21b7c814e89efe/scikit_learn-1.8.0-cp311-cp311-win_arm64.whl", hash = "sha256:2838551e011a64e3053ad7618dda9310175f7515f1742fa2d756f7c874c05961", size = 7688451, upload-time = "2025-12-10T07:07:49.873Z" },
+    { url = "https://files.pythonhosted.org/packages/90/74/e6a7cc4b820e95cc38cf36cd74d5aa2b42e8ffc2d21fe5a9a9c45c1c7630/scikit_learn-1.8.0-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:5fb63362b5a7ddab88e52b6dbb47dac3fd7dafeee740dc6c8d8a446ddedade8e", size = 8548242, upload-time = "2025-12-10T07:07:51.568Z" },
+    { url = "https://files.pythonhosted.org/packages/49/d8/9be608c6024d021041c7f0b3928d4749a706f4e2c3832bbede4fb4f58c95/scikit_learn-1.8.0-cp312-cp312-macosx_12_0_arm64.whl", hash = "sha256:5025ce924beccb28298246e589c691fe1b8c1c96507e6d27d12c5fadd85bfd76", size = 8079075, upload-time = "2025-12-10T07:07:53.697Z" },
+    { url = "https://files.pythonhosted.org/packages/dd/47/f187b4636ff80cc63f21cd40b7b2d177134acaa10f6bb73746130ee8c2e5/scikit_learn-1.8.0-cp312-cp312-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:4496bb2cf7a43ce1a2d7524a79e40bc5da45cf598dbf9545b7e8316ccba47bb4", size = 8660492, upload-time = "2025-12-10T07:07:55.574Z" },
+    { url = "https://files.pythonhosted.org/packages/97/74/b7a304feb2b49df9fafa9382d4d09061a96ee9a9449a7cbea7988dda0828/scikit_learn-1.8.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:a0bcfe4d0d14aec44921545fd2af2338c7471de9cb701f1da4c9d85906ab847a", size = 8931904, upload-time = "2025-12-10T07:07:57.666Z" },
+    { url = "https://files.pythonhosted.org/packages/9f/c4/0ab22726a04ede56f689476b760f98f8f46607caecff993017ac1b64aa5d/scikit_learn-1.8.0-cp312-cp312-win_amd64.whl", hash = "sha256:35c007dedb2ffe38fe3ee7d201ebac4a2deccd2408e8621d53067733e3c74809", size = 8019359, upload-time = "2025-12-10T07:07:59.838Z" },
+    { url = "https://files.pythonhosted.org/packages/24/90/344a67811cfd561d7335c1b96ca21455e7e472d281c3c279c4d3f2300236/scikit_learn-1.8.0-cp312-cp312-win_arm64.whl", hash = "sha256:8c497fff237d7b4e07e9ef1a640887fa4fb765647f86fbe00f969ff6280ce2bb", size = 7641898, upload-time = "2025-12-10T07:08:01.36Z" },
+    { url = "https://files.pythonhosted.org/packages/03/aa/e22e0768512ce9255eba34775be2e85c2048da73da1193e841707f8f039c/scikit_learn-1.8.0-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:0d6ae97234d5d7079dc0040990a6f7aeb97cb7fa7e8945f1999a429b23569e0a", size = 8513770, upload-time = "2025-12-10T07:08:03.251Z" },
+    { url = "https://files.pythonhosted.org/packages/58/37/31b83b2594105f61a381fc74ca19e8780ee923be2d496fcd8d2e1147bd99/scikit_learn-1.8.0-cp313-cp313-macosx_12_0_arm64.whl", hash = "sha256:edec98c5e7c128328124a029bceb09eda2d526997780fef8d65e9a69eead963e", size = 8044458, upload-time = "2025-12-10T07:08:05.336Z" },
+    { url = "https://files.pythonhosted.org/packages/2d/5a/3f1caed8765f33eabb723596666da4ebbf43d11e96550fb18bdec42b467b/scikit_learn-1.8.0-cp313-cp313-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:74b66d8689d52ed04c271e1329f0c61635bcaf5b926db9b12d58914cdc01fe57", size = 8610341, upload-time = "2025-12-10T07:08:07.732Z" },
+    { url = "https://files.pythonhosted.org/packages/38/cf/06896db3f71c75902a8e9943b444a56e727418f6b4b4a90c98c934f51ed4/scikit_learn-1.8.0-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:8fdf95767f989b0cfedb85f7ed8ca215d4be728031f56ff5a519ee1e3276dc2e", size = 8900022, upload-time = "2025-12-10T07:08:09.862Z" },
+    { url = "https://files.pythonhosted.org/packages/1c/f9/9b7563caf3ec8873e17a31401858efab6b39a882daf6c1bfa88879c0aa11/scikit_learn-1.8.0-cp313-cp313-win_amd64.whl", hash = "sha256:2de443b9373b3b615aec1bb57f9baa6bb3a9bd093f1269ba95c17d870422b271", size = 7989409, upload-time = "2025-12-10T07:08:12.028Z" },
+    { url = "https://files.pythonhosted.org/packages/49/bd/1f4001503650e72c4f6009ac0c4413cb17d2d601cef6f71c0453da2732fc/scikit_learn-1.8.0-cp313-cp313-win_arm64.whl", hash = "sha256:eddde82a035681427cbedded4e6eff5e57fa59216c2e3e90b10b19ab1d0a65c3", size = 7619760, upload-time = "2025-12-10T07:08:13.688Z" },
+    { url = "https://files.pythonhosted.org/packages/d2/7d/a630359fc9dcc95496588c8d8e3245cc8fd81980251079bc09c70d41d951/scikit_learn-1.8.0-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:7cc267b6108f0a1499a734167282c00c4ebf61328566b55ef262d48e9849c735", size = 8826045, upload-time = "2025-12-10T07:08:15.215Z" },
+    { url = "https://files.pythonhosted.org/packages/cc/56/a0c86f6930cfcd1c7054a2bc417e26960bb88d32444fe7f71d5c2cfae891/scikit_learn-1.8.0-cp313-cp313t-macosx_12_0_arm64.whl", hash = "sha256:fe1c011a640a9f0791146011dfd3c7d9669785f9fed2b2a5f9e207536cf5c2fd", size = 8420324, upload-time = "2025-12-10T07:08:17.561Z" },
+    { url = "https://files.pythonhosted.org/packages/46/1e/05962ea1cebc1cf3876667ecb14c283ef755bf409993c5946ade3b77e303/scikit_learn-1.8.0-cp313-cp313t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:72358cce49465d140cc4e7792015bb1f0296a9742d5622c67e31399b75468b9e", size = 8680651, upload-time = "2025-12-10T07:08:19.952Z" },
+    { url = "https://files.pythonhosted.org/packages/fe/56/a85473cd75f200c9759e3a5f0bcab2d116c92a8a02ee08ccd73b870f8bb4/scikit_learn-1.8.0-cp313-cp313t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:80832434a6cc114f5219211eec13dcbc16c2bac0e31ef64c6d346cde3cf054cb", size = 8925045, upload-time = "2025-12-10T07:08:22.11Z" },
+    { url = "https://files.pythonhosted.org/packages/cc/b7/64d8cfa896c64435ae57f4917a548d7ac7a44762ff9802f75a79b77cb633/scikit_learn-1.8.0-cp313-cp313t-win_amd64.whl", hash = "sha256:ee787491dbfe082d9c3013f01f5991658b0f38aa8177e4cd4bf434c58f551702", size = 8507994, upload-time = "2025-12-10T07:08:23.943Z" },
+    { url = "https://files.pythonhosted.org/packages/5e/37/e192ea709551799379958b4c4771ec507347027bb7c942662c7fbeba31cb/scikit_learn-1.8.0-cp313-cp313t-win_arm64.whl", hash = "sha256:bf97c10a3f5a7543f9b88cbf488d33d175e9146115a451ae34568597ba33dcde", size = 7869518, upload-time = "2025-12-10T07:08:25.71Z" },
+    { url = "https://files.pythonhosted.org/packages/24/05/1af2c186174cc92dcab2233f327336058c077d38f6fe2aceb08e6ab4d509/scikit_learn-1.8.0-cp314-cp314-macosx_10_15_x86_64.whl", hash = "sha256:c22a2da7a198c28dd1a6e1136f19c830beab7fdca5b3e5c8bba8394f8a5c45b3", size = 8528667, upload-time = "2025-12-10T07:08:27.541Z" },
+    { url = "https://files.pythonhosted.org/packages/a8/25/01c0af38fe969473fb292bba9dc2b8f9b451f3112ff242c647fee3d0dfe7/scikit_learn-1.8.0-cp314-cp314-macosx_12_0_arm64.whl", hash = "sha256:6b595b07a03069a2b1740dc08c2299993850ea81cce4fe19b2421e0c970de6b7", size = 8066524, upload-time = "2025-12-10T07:08:29.822Z" },
+    { url = "https://files.pythonhosted.org/packages/be/ce/a0623350aa0b68647333940ee46fe45086c6060ec604874e38e9ab7d8e6c/scikit_learn-1.8.0-cp314-cp314-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:29ffc74089f3d5e87dfca4c2c8450f88bdc61b0fc6ed5d267f3988f19a1309f6", size = 8657133, upload-time = "2025-12-10T07:08:31.865Z" },
+    { url = "https://files.pythonhosted.org/packages/b8/cb/861b41341d6f1245e6ca80b1c1a8c4dfce43255b03df034429089ca2a2c5/scikit_learn-1.8.0-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:fb65db5d7531bccf3a4f6bec3462223bea71384e2cda41da0f10b7c292b9e7c4", size = 8923223, upload-time = "2025-12-10T07:08:34.166Z" },
+    { url = "https://files.pythonhosted.org/packages/76/18/a8def8f91b18cd1ba6e05dbe02540168cb24d47e8dcf69e8d00b7da42a08/scikit_learn-1.8.0-cp314-cp314-win_amd64.whl", hash = "sha256:56079a99c20d230e873ea40753102102734c5953366972a71d5cb39a32bc40c6", size = 8096518, upload-time = "2025-12-10T07:08:36.339Z" },
+    { url = "https://files.pythonhosted.org/packages/d1/77/482076a678458307f0deb44e29891d6022617b2a64c840c725495bee343f/scikit_learn-1.8.0-cp314-cp314-win_arm64.whl", hash = "sha256:3bad7565bc9cf37ce19a7c0d107742b320c1285df7aab1a6e2d28780df167242", size = 7754546, upload-time = "2025-12-10T07:08:38.128Z" },
+    { url = "https://files.pythonhosted.org/packages/2d/d1/ef294ca754826daa043b2a104e59960abfab4cf653891037d19dd5b6f3cf/scikit_learn-1.8.0-cp314-cp314t-macosx_10_15_x86_64.whl", hash = "sha256:4511be56637e46c25721e83d1a9cea9614e7badc7040c4d573d75fbe257d6fd7", size = 8848305, upload-time = "2025-12-10T07:08:41.013Z" },
+    { url = "https://files.pythonhosted.org/packages/5b/e2/b1f8b05138ee813b8e1a4149f2f0d289547e60851fd1bb268886915adbda/scikit_learn-1.8.0-cp314-cp314t-macosx_12_0_arm64.whl", hash = "sha256:a69525355a641bf8ef136a7fa447672fb54fe8d60cab5538d9eb7c6438543fb9", size = 8432257, upload-time = "2025-12-10T07:08:42.873Z" },
+    { url = "https://files.pythonhosted.org/packages/26/11/c32b2138a85dcb0c99f6afd13a70a951bfdff8a6ab42d8160522542fb647/scikit_learn-1.8.0-cp314-cp314t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:c2656924ec73e5939c76ac4c8b026fc203b83d8900362eb2599d8aee80e4880f", size = 8678673, upload-time = "2025-12-10T07:08:45.362Z" },
+    { url = "https://files.pythonhosted.org/packages/c7/57/51f2384575bdec454f4fe4e7a919d696c9ebce914590abf3e52d47607ab8/scikit_learn-1.8.0-cp314-cp314t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:15fc3b5d19cc2be65404786857f2e13c70c83dd4782676dd6814e3b89dc8f5b9", size = 8922467, upload-time = "2025-12-10T07:08:47.408Z" },
+    { url = "https://files.pythonhosted.org/packages/35/4d/748c9e2872637a57981a04adc038dacaa16ba8ca887b23e34953f0b3f742/scikit_learn-1.8.0-cp314-cp314t-win_amd64.whl", hash = "sha256:00d6f1d66fbcf4eba6e356e1420d33cc06c70a45bb1363cd6f6a8e4ebbbdece2", size = 8774395, upload-time = "2025-12-10T07:08:49.337Z" },
+    { url = "https://files.pythonhosted.org/packages/60/22/d7b2ebe4704a5e50790ba089d5c2ae308ab6bb852719e6c3bd4f04c3a363/scikit_learn-1.8.0-cp314-cp314t-win_arm64.whl", hash = "sha256:f28dd15c6bb0b66ba09728cf09fd8736c304be29409bd8445a080c1280619e8c", size = 8002647, upload-time = "2025-12-10T07:08:51.601Z" },
+]
+
+[[package]]
+name = "scipy"
+version = "1.17.1"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "numpy", marker = "python_full_version >= '3.14'" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/7a/97/5a3609c4f8d58b039179648e62dd220f89864f56f7357f5d4f45c29eb2cc/scipy-1.17.1.tar.gz", hash = "sha256:95d8e012d8cb8816c226aef832200b1d45109ed4464303e997c5b13122b297c0", size = 30573822, upload-time = "2026-02-23T00:26:24.851Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/df/75/b4ce781849931fef6fd529afa6b63711d5a733065722d0c3e2724af9e40a/scipy-1.17.1-cp311-cp311-macosx_10_14_x86_64.whl", hash = "sha256:1f95b894f13729334fb990162e911c9e5dc1ab390c58aa6cbecb389c5b5e28ec", size = 31613675, upload-time = "2026-02-23T00:16:00.13Z" },
+    { url = "https://files.pythonhosted.org/packages/f7/58/bccc2861b305abdd1b8663d6130c0b3d7cc22e8d86663edbc8401bfd40d4/scipy-1.17.1-cp311-cp311-macosx_12_0_arm64.whl", hash = "sha256:e18f12c6b0bc5a592ed23d3f7b891f68fd7f8241d69b7883769eb5d5dfb52696", size = 28162057, upload-time = "2026-02-23T00:16:09.456Z" },
+    { url = "https://files.pythonhosted.org/packages/6d/ee/18146b7757ed4976276b9c9819108adbc73c5aad636e5353e20746b73069/scipy-1.17.1-cp311-cp311-macosx_14_0_arm64.whl", hash = "sha256:a3472cfbca0a54177d0faa68f697d8ba4c80bbdc19908c3465556d9f7efce9ee", size = 20334032, upload-time = "2026-02-23T00:16:17.358Z" },
+    { url = "https://files.pythonhosted.org/packages/ec/e6/cef1cf3557f0c54954198554a10016b6a03b2ec9e22a4e1df734936bd99c/scipy-1.17.1-cp311-cp311-macosx_14_0_x86_64.whl", hash = "sha256:766e0dc5a616d026a3a1cffa379af959671729083882f50307e18175797b3dfd", size = 22709533, upload-time = "2026-02-23T00:16:25.791Z" },
+    { url = "https://files.pythonhosted.org/packages/4d/60/8804678875fc59362b0fb759ab3ecce1f09c10a735680318ac30da8cd76b/scipy-1.17.1-cp311-cp311-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:744b2bf3640d907b79f3fd7874efe432d1cf171ee721243e350f55234b4cec4c", size = 33062057, upload-time = "2026-02-23T00:16:36.931Z" },
+    { url = "https://files.pythonhosted.org/packages/09/7d/af933f0f6e0767995b4e2d705a0665e454d1c19402aa7e895de3951ebb04/scipy-1.17.1-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:43af8d1f3bea642559019edfe64e9b11192a8978efbd1539d7bc2aaa23d92de4", size = 35349300, upload-time = "2026-02-23T00:16:49.108Z" },
+    { url = "https://files.pythonhosted.org/packages/b4/3d/7ccbbdcbb54c8fdc20d3b6930137c782a163fa626f0aef920349873421ba/scipy-1.17.1-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:cd96a1898c0a47be4520327e01f874acfd61fb48a9420f8aa9f6483412ffa444", size = 35127333, upload-time = "2026-02-23T00:17:01.293Z" },
+    { url = "https://files.pythonhosted.org/packages/e8/19/f926cb11c42b15ba08e3a71e376d816ac08614f769b4f47e06c3580c836a/scipy-1.17.1-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:4eb6c25dd62ee8d5edf68a8e1c171dd71c292fdae95d8aeb3dd7d7de4c364082", size = 37741314, upload-time = "2026-02-23T00:17:12.576Z" },
+    { url = "https://files.pythonhosted.org/packages/95/da/0d1df507cf574b3f224ccc3d45244c9a1d732c81dcb26b1e8a766ae271a8/scipy-1.17.1-cp311-cp311-win_amd64.whl", hash = "sha256:d30e57c72013c2a4fe441c2fcb8e77b14e152ad48b5464858e07e2ad9fbfceff", size = 36607512, upload-time = "2026-02-23T00:17:23.424Z" },
+    { url = "https://files.pythonhosted.org/packages/68/7f/bdd79ceaad24b671543ffe0ef61ed8e659440eb683b66f033454dcee90eb/scipy-1.17.1-cp311-cp311-win_arm64.whl", hash = "sha256:9ecb4efb1cd6e8c4afea0daa91a87fbddbce1b99d2895d151596716c0b2e859d", size = 24599248, upload-time = "2026-02-23T00:17:34.561Z" },
+    { url = "https://files.pythonhosted.org/packages/35/48/b992b488d6f299dbe3f11a20b24d3dda3d46f1a635ede1c46b5b17a7b163/scipy-1.17.1-cp312-cp312-macosx_10_14_x86_64.whl", hash = "sha256:35c3a56d2ef83efc372eaec584314bd0ef2e2f0d2adb21c55e6ad5b344c0dcb8", size = 31610954, upload-time = "2026-02-23T00:17:49.855Z" },
+    { url = "https://files.pythonhosted.org/packages/b2/02/cf107b01494c19dc100f1d0b7ac3cc08666e96ba2d64db7626066cee895e/scipy-1.17.1-cp312-cp312-macosx_12_0_arm64.whl", hash = "sha256:fcb310ddb270a06114bb64bbe53c94926b943f5b7f0842194d585c65eb4edd76", size = 28172662, upload-time = "2026-02-23T00:18:01.64Z" },
+    { url = "https://files.pythonhosted.org/packages/cf/a9/599c28631bad314d219cf9ffd40e985b24d603fc8a2f4ccc5ae8419a535b/scipy-1.17.1-cp312-cp312-macosx_14_0_arm64.whl", hash = "sha256:cc90d2e9c7e5c7f1a482c9875007c095c3194b1cfedca3c2f3291cdc2bc7c086", size = 20344366, upload-time = "2026-02-23T00:18:12.015Z" },
+    { url = "https://files.pythonhosted.org/packages/35/f5/906eda513271c8deb5af284e5ef0206d17a96239af79f9fa0aebfe0e36b4/scipy-1.17.1-cp312-cp312-macosx_14_0_x86_64.whl", hash = "sha256:c80be5ede8f3f8eded4eff73cc99a25c388ce98e555b17d31da05287015ffa5b", size = 22704017, upload-time = "2026-02-23T00:18:21.502Z" },
+    { url = "https://files.pythonhosted.org/packages/da/34/16f10e3042d2f1d6b66e0428308ab52224b6a23049cb2f5c1756f713815f/scipy-1.17.1-cp312-cp312-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:e19ebea31758fac5893a2ac360fedd00116cbb7628e650842a6691ba7ca28a21", size = 32927842, upload-time = "2026-02-23T00:18:35.367Z" },
+    { url = "https://files.pythonhosted.org/packages/01/8e/1e35281b8ab6d5d72ebe9911edcdffa3f36b04ed9d51dec6dd140396e220/scipy-1.17.1-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:02ae3b274fde71c5e92ac4d54bc06c42d80e399fec704383dcd99b301df37458", size = 35235890, upload-time = "2026-02-23T00:18:49.188Z" },
+    { url = "https://files.pythonhosted.org/packages/c5/5c/9d7f4c88bea6e0d5a4f1bc0506a53a00e9fcb198de372bfe4d3652cef482/scipy-1.17.1-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:8a604bae87c6195d8b1045eddece0514d041604b14f2727bbc2b3020172045eb", size = 35003557, upload-time = "2026-02-23T00:18:54.74Z" },
+    { url = "https://files.pythonhosted.org/packages/65/94/7698add8f276dbab7a9de9fb6b0e02fc13ee61d51c7c3f85ac28b65e1239/scipy-1.17.1-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:f590cd684941912d10becc07325a3eeb77886fe981415660d9265c4c418d0bea", size = 37625856, upload-time = "2026-02-23T00:19:00.307Z" },
+    { url = "https://files.pythonhosted.org/packages/a2/84/dc08d77fbf3d87d3ee27f6a0c6dcce1de5829a64f2eae85a0ecc1f0daa73/scipy-1.17.1-cp312-cp312-win_amd64.whl", hash = "sha256:41b71f4a3a4cab9d366cd9065b288efc4d4f3c0b37a91a8e0947fb5bd7f31d87", size = 36549682, upload-time = "2026-02-23T00:19:07.67Z" },
+    { url = "https://files.pythonhosted.org/packages/bc/98/fe9ae9ffb3b54b62559f52dedaebe204b408db8109a8c66fdd04869e6424/scipy-1.17.1-cp312-cp312-win_arm64.whl", hash = "sha256:f4115102802df98b2b0db3cce5cb9b92572633a1197c77b7553e5203f284a5b3", size = 24547340, upload-time = "2026-02-23T00:19:12.024Z" },
+    { url = "https://files.pythonhosted.org/packages/76/27/07ee1b57b65e92645f219b37148a7e7928b82e2b5dbeccecb4dff7c64f0b/scipy-1.17.1-cp313-cp313-macosx_10_14_x86_64.whl", hash = "sha256:5e3c5c011904115f88a39308379c17f91546f77c1667cea98739fe0fccea804c", size = 31590199, upload-time = "2026-02-23T00:19:17.192Z" },
+    { url = "https://files.pythonhosted.org/packages/ec/ae/db19f8ab842e9b724bf5dbb7db29302a91f1e55bc4d04b1025d6d605a2c5/scipy-1.17.1-cp313-cp313-macosx_12_0_arm64.whl", hash = "sha256:6fac755ca3d2c3edcb22f479fceaa241704111414831ddd3bc6056e18516892f", size = 28154001, upload-time = "2026-02-23T00:19:22.241Z" },
+    { url = "https://files.pythonhosted.org/packages/5b/58/3ce96251560107b381cbd6e8413c483bbb1228a6b919fa8652b0d4090e7f/scipy-1.17.1-cp313-cp313-macosx_14_0_arm64.whl", hash = "sha256:7ff200bf9d24f2e4d5dc6ee8c3ac64d739d3a89e2326ba68aaf6c4a2b838fd7d", size = 20325719, upload-time = "2026-02-23T00:19:26.329Z" },
+    { url = "https://files.pythonhosted.org/packages/b2/83/15087d945e0e4d48ce2377498abf5ad171ae013232ae31d06f336e64c999/scipy-1.17.1-cp313-cp313-macosx_14_0_x86_64.whl", hash = "sha256:4b400bdc6f79fa02a4d86640310dde87a21fba0c979efff5248908c6f15fad1b", size = 22683595, upload-time = "2026-02-23T00:19:30.304Z" },
+    { url = "https://files.pythonhosted.org/packages/b4/e0/e58fbde4a1a594c8be8114eb4aac1a55bcd6587047efc18a61eb1f5c0d30/scipy-1.17.1-cp313-cp313-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:2b64ca7d4aee0102a97f3ba22124052b4bd2152522355073580bf4845e2550b6", size = 32896429, upload-time = "2026-02-23T00:19:35.536Z" },
+    { url = "https://files.pythonhosted.org/packages/f5/5f/f17563f28ff03c7b6799c50d01d5d856a1d55f2676f537ca8d28c7f627cd/scipy-1.17.1-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:581b2264fc0aa555f3f435a5944da7504ea3a065d7029ad60e7c3d1ae09c5464", size = 35203952, upload-time = "2026-02-23T00:19:42.259Z" },
+    { url = "https://files.pythonhosted.org/packages/8d/a5/9afd17de24f657fdfe4df9a3f1ea049b39aef7c06000c13db1530d81ccca/scipy-1.17.1-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:beeda3d4ae615106d7094f7e7cef6218392e4465cc95d25f900bebabfded0950", size = 34979063, upload-time = "2026-02-23T00:19:47.547Z" },
+    { url = "https://files.pythonhosted.org/packages/8b/13/88b1d2384b424bf7c924f2038c1c409f8d88bb2a8d49d097861dd64a57b2/scipy-1.17.1-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:6609bc224e9568f65064cfa72edc0f24ee6655b47575954ec6339534b2798369", size = 37598449, upload-time = "2026-02-23T00:19:53.238Z" },
+    { url = "https://files.pythonhosted.org/packages/35/e5/d6d0e51fc888f692a35134336866341c08655d92614f492c6860dc45bb2c/scipy-1.17.1-cp313-cp313-win_amd64.whl", hash = "sha256:37425bc9175607b0268f493d79a292c39f9d001a357bebb6b88fdfaff13f6448", size = 36510943, upload-time = "2026-02-23T00:20:50.89Z" },
+    { url = "https://files.pythonhosted.org/packages/2a/fd/3be73c564e2a01e690e19cc618811540ba5354c67c8680dce3281123fb79/scipy-1.17.1-cp313-cp313-win_arm64.whl", hash = "sha256:5cf36e801231b6a2059bf354720274b7558746f3b1a4efb43fcf557ccd484a87", size = 24545621, upload-time = "2026-02-23T00:20:55.871Z" },
+    { url = "https://files.pythonhosted.org/packages/6f/6b/17787db8b8114933a66f9dcc479a8272e4b4da75fe03b0c282f7b0ade8cd/scipy-1.17.1-cp313-cp313t-macosx_10_14_x86_64.whl", hash = "sha256:d59c30000a16d8edc7e64152e30220bfbd724c9bbb08368c054e24c651314f0a", size = 31936708, upload-time = "2026-02-23T00:19:58.694Z" },
+    { url = "https://files.pythonhosted.org/packages/38/2e/524405c2b6392765ab1e2b722a41d5da33dc5c7b7278184a8ad29b6cb206/scipy-1.17.1-cp313-cp313t-macosx_12_0_arm64.whl", hash = "sha256:010f4333c96c9bb1a4516269e33cb5917b08ef2166d5556ca2fd9f082a9e6ea0", size = 28570135, upload-time = "2026-02-23T00:20:03.934Z" },
+    { url = "https://files.pythonhosted.org/packages/fd/c3/5bd7199f4ea8556c0c8e39f04ccb014ac37d1468e6cfa6a95c6b3562b76e/scipy-1.17.1-cp313-cp313t-macosx_14_0_arm64.whl", hash = "sha256:2ceb2d3e01c5f1d83c4189737a42d9cb2fc38a6eeed225e7515eef71ad301dce", size = 20741977, upload-time = "2026-02-23T00:20:07.935Z" },
+    { url = "https://files.pythonhosted.org/packages/d9/b8/8ccd9b766ad14c78386599708eb745f6b44f08400a5fd0ade7cf89b6fc93/scipy-1.17.1-cp313-cp313t-macosx_14_0_x86_64.whl", hash = "sha256:844e165636711ef41f80b4103ed234181646b98a53c8f05da12ca5ca289134f6", size = 23029601, upload-time = "2026-02-23T00:20:12.161Z" },
+    { url = "https://files.pythonhosted.org/packages/6d/a0/3cb6f4d2fb3e17428ad2880333cac878909ad1a89f678527b5328b93c1d4/scipy-1.17.1-cp313-cp313t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:158dd96d2207e21c966063e1635b1063cd7787b627b6f07305315dd73d9c679e", size = 33019667, upload-time = "2026-02-23T00:20:17.208Z" },
+    { url = "https://files.pythonhosted.org/packages/f3/c3/2d834a5ac7bf3a0c806ad1508efc02dda3c8c61472a56132d7894c312dea/scipy-1.17.1-cp313-cp313t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:74cbb80d93260fe2ffa334efa24cb8f2f0f622a9b9febf8b483c0b865bfb3475", size = 35264159, upload-time = "2026-02-23T00:20:23.087Z" },
+    { url = "https://files.pythonhosted.org/packages/4d/77/d3ed4becfdbd217c52062fafe35a72388d1bd82c2d0ba5ca19d6fcc93e11/scipy-1.17.1-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:dbc12c9f3d185f5c737d801da555fb74b3dcfa1a50b66a1a93e09190f41fab50", size = 35102771, upload-time = "2026-02-23T00:20:28.636Z" },
+    { url = "https://files.pythonhosted.org/packages/bd/12/d19da97efde68ca1ee5538bb261d5d2c062f0c055575128f11a2730e3ac1/scipy-1.17.1-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:94055a11dfebe37c656e70317e1996dc197e1a15bbcc351bcdd4610e128fe1ca", size = 37665910, upload-time = "2026-02-23T00:20:34.743Z" },
+    { url = "https://files.pythonhosted.org/packages/06/1c/1172a88d507a4baaf72c5a09bb6c018fe2ae0ab622e5830b703a46cc9e44/scipy-1.17.1-cp313-cp313t-win_amd64.whl", hash = "sha256:e30bdeaa5deed6bc27b4cc490823cd0347d7dae09119b8803ae576ea0ce52e4c", size = 36562980, upload-time = "2026-02-23T00:20:40.575Z" },
+    { url = "https://files.pythonhosted.org/packages/70/b0/eb757336e5a76dfa7911f63252e3b7d1de00935d7705cf772db5b45ec238/scipy-1.17.1-cp313-cp313t-win_arm64.whl", hash = "sha256:a720477885a9d2411f94a93d16f9d89bad0f28ca23c3f8daa521e2dcc3f44d49", size = 24856543, upload-time = "2026-02-23T00:20:45.313Z" },
+    { url = "https://files.pythonhosted.org/packages/cf/83/333afb452af6f0fd70414dc04f898647ee1423979ce02efa75c3b0f2c28e/scipy-1.17.1-cp314-cp314-macosx_10_14_x86_64.whl", hash = "sha256:a48a72c77a310327f6a3a920092fa2b8fd03d7deaa60f093038f22d98e096717", size = 31584510, upload-time = "2026-02-23T00:21:01.015Z" },
+    { url = "https://files.pythonhosted.org/packages/ed/a6/d05a85fd51daeb2e4ea71d102f15b34fedca8e931af02594193ae4fd25f7/scipy-1.17.1-cp314-cp314-macosx_12_0_arm64.whl", hash = "sha256:45abad819184f07240d8a696117a7aacd39787af9e0b719d00285549ed19a1e9", size = 28170131, upload-time = "2026-02-23T00:21:05.888Z" },
+    { url = "https://files.pythonhosted.org/packages/db/7b/8624a203326675d7746a254083a187398090a179335b2e4a20e2ddc46e83/scipy-1.17.1-cp314-cp314-macosx_14_0_arm64.whl", hash = "sha256:3fd1fcdab3ea951b610dc4cef356d416d5802991e7e32b5254828d342f7b7e0b", size = 20342032, upload-time = "2026-02-23T00:21:09.904Z" },
+    { url = "https://files.pythonhosted.org/packages/c9/35/2c342897c00775d688d8ff3987aced3426858fd89d5a0e26e020b660b301/scipy-1.17.1-cp314-cp314-macosx_14_0_x86_64.whl", hash = "sha256:7bdf2da170b67fdf10bca777614b1c7d96ae3ca5794fd9587dce41eb2966e866", size = 22678766, upload-time = "2026-02-23T00:21:14.313Z" },
+    { url = "https://files.pythonhosted.org/packages/ef/f2/7cdb8eb308a1a6ae1e19f945913c82c23c0c442a462a46480ce487fdc0ac/scipy-1.17.1-cp314-cp314-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:adb2642e060a6549c343603a3851ba76ef0b74cc8c079a9a58121c7ec9fe2350", size = 32957007, upload-time = "2026-02-23T00:21:19.663Z" },
+    { url = "https://files.pythonhosted.org/packages/0b/2e/7eea398450457ecb54e18e9d10110993fa65561c4f3add5e8eccd2b9cd41/scipy-1.17.1-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:eee2cfda04c00a857206a4330f0c5e3e56535494e30ca445eb19ec624ae75118", size = 35221333, upload-time = "2026-02-23T00:21:25.278Z" },
+    { url = "https://files.pythonhosted.org/packages/d9/77/5b8509d03b77f093a0d52e606d3c4f79e8b06d1d38c441dacb1e26cacf46/scipy-1.17.1-cp314-cp314-musllinux_1_2_aarch64.whl", hash = "sha256:d2650c1fb97e184d12d8ba010493ee7b322864f7d3d00d3f9bb97d9c21de4068", size = 35042066, upload-time = "2026-02-23T00:21:31.358Z" },
+    { url = "https://files.pythonhosted.org/packages/f9/df/18f80fb99df40b4070328d5ae5c596f2f00fffb50167e31439e932f29e7d/scipy-1.17.1-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:08b900519463543aa604a06bec02461558a6e1cef8fdbb8098f77a48a83c8118", size = 37612763, upload-time = "2026-02-23T00:21:37.247Z" },
+    { url = "https://files.pythonhosted.org/packages/4b/39/f0e8ea762a764a9dc52aa7dabcfad51a354819de1f0d4652b6a1122424d6/scipy-1.17.1-cp314-cp314-win_amd64.whl", hash = "sha256:3877ac408e14da24a6196de0ddcace62092bfc12a83823e92e49e40747e52c19", size = 37290984, upload-time = "2026-02-23T00:22:35.023Z" },
+    { url = "https://files.pythonhosted.org/packages/7c/56/fe201e3b0f93d1a8bcf75d3379affd228a63d7e2d80ab45467a74b494947/scipy-1.17.1-cp314-cp314-win_arm64.whl", hash = "sha256:f8885db0bc2bffa59d5c1b72fad7a6a92d3e80e7257f967dd81abb553a90d293", size = 25192877, upload-time = "2026-02-23T00:22:39.798Z" },
+    { url = "https://files.pythonhosted.org/packages/96/ad/f8c414e121f82e02d76f310f16db9899c4fcde36710329502a6b2a3c0392/scipy-1.17.1-cp314-cp314t-macosx_10_14_x86_64.whl", hash = "sha256:1cc682cea2ae55524432f3cdff9e9a3be743d52a7443d0cba9017c23c87ae2f6", size = 31949750, upload-time = "2026-02-23T00:21:42.289Z" },
+    { url = "https://files.pythonhosted.org/packages/7c/b0/c741e8865d61b67c81e255f4f0a832846c064e426636cd7de84e74d209be/scipy-1.17.1-cp314-cp314t-macosx_12_0_arm64.whl", hash = "sha256:2040ad4d1795a0ae89bfc7e8429677f365d45aa9fd5e4587cf1ea737f927b4a1", size = 28585858, upload-time = "2026-02-23T00:21:47.706Z" },
+    { url = "https://files.pythonhosted.org/packages/ed/1b/3985219c6177866628fa7c2595bfd23f193ceebbe472c98a08824b9466ff/scipy-1.17.1-cp314-cp314t-macosx_14_0_arm64.whl", hash = "sha256:131f5aaea57602008f9822e2115029b55d4b5f7c070287699fe45c661d051e39", size = 20757723, upload-time = "2026-02-23T00:21:52.039Z" },
+    { url = "https://files.pythonhosted.org/packages/c0/19/2a04aa25050d656d6f7b9e7b685cc83d6957fb101665bfd9369ca6534563/scipy-1.17.1-cp314-cp314t-macosx_14_0_x86_64.whl", hash = "sha256:9cdc1a2fcfd5c52cfb3045feb399f7b3ce822abdde3a193a6b9a60b3cb5854ca", size = 23043098, upload-time = "2026-02-23T00:21:56.185Z" },
+    { url = "https://files.pythonhosted.org/packages/86/f1/3383beb9b5d0dbddd030335bf8a8b32d4317185efe495374f134d8be6cce/scipy-1.17.1-cp314-cp314t-manylinux_2_27_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:6e3dcd57ab780c741fde8dc68619de988b966db759a3c3152e8e9142c26295ad", size = 33030397, upload-time = "2026-02-23T00:22:01.404Z" },
+    { url = "https://files.pythonhosted.org/packages/41/68/8f21e8a65a5a03f25a79165ec9d2b28c00e66dc80546cf5eb803aeeff35b/scipy-1.17.1-cp314-cp314t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:a9956e4d4f4a301ebf6cde39850333a6b6110799d470dbbb1e25326ac447f52a", size = 35281163, upload-time = "2026-02-23T00:22:07.024Z" },
+    { url = "https://files.pythonhosted.org/packages/84/8d/c8a5e19479554007a5632ed7529e665c315ae7492b4f946b0deb39870e39/scipy-1.17.1-cp314-cp314t-musllinux_1_2_aarch64.whl", hash = "sha256:a4328d245944d09fd639771de275701ccadf5f781ba0ff092ad141e017eccda4", size = 35116291, upload-time = "2026-02-23T00:22:12.585Z" },
+    { url = "https://files.pythonhosted.org/packages/52/52/e57eceff0e342a1f50e274264ed47497b59e6a4e3118808ee58ddda7b74a/scipy-1.17.1-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:a77cbd07b940d326d39a1d1b37817e2ee4d79cb30e7338f3d0cddffae70fcaa2", size = 37682317, upload-time = "2026-02-23T00:22:18.513Z" },
+    { url = "https://files.pythonhosted.org/packages/11/2f/b29eafe4a3fbc3d6de9662b36e028d5f039e72d345e05c250e121a230dd4/scipy-1.17.1-cp314-cp314t-win_amd64.whl", hash = "sha256:eb092099205ef62cd1782b006658db09e2fed75bffcae7cc0d44052d8aa0f484", size = 37345327, upload-time = "2026-02-23T00:22:24.442Z" },
+    { url = "https://files.pythonhosted.org/packages/07/39/338d9219c4e87f3e708f18857ecd24d22a0c3094752393319553096b98af/scipy-1.17.1-cp314-cp314t-win_arm64.whl", hash = "sha256:200e1050faffacc162be6a486a984a0497866ec54149a01270adc8a59b7c7d21", size = 25489165, upload-time = "2026-02-23T00:22:29.563Z" },
+]
+
+[[package]]
+name = "sentence-transformers"
+version = "5.2.3"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "huggingface-hub", marker = "python_full_version >= '3.14'" },
+    { name = "numpy", marker = "python_full_version >= '3.14'" },
+    { name = "scikit-learn", marker = "python_full_version >= '3.14'" },
+    { name = "scipy", marker = "python_full_version >= '3.14'" },
+    { name = "torch", marker = "python_full_version >= '3.14'" },
+    { name = "tqdm", marker = "python_full_version >= '3.14'" },
+    { name = "transformers", marker = "python_full_version >= '3.14'" },
+    { name = "typing-extensions", marker = "python_full_version >= '3.14'" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/5b/30/21664028fc0776eb1ca024879480bbbab36f02923a8ff9e4cae5a150fa35/sentence_transformers-5.2.3.tar.gz", hash = "sha256:3cd3044e1f3fe859b6a1b66336aac502eaae5d3dd7d5c8fc237f37fbf58137c7", size = 381623, upload-time = "2026-02-17T14:05:20.238Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/46/9f/dba4b3e18ebbe1eaa29d9f1764fbc7da0cd91937b83f2b7928d15c5d2d36/sentence_transformers-5.2.3-py3-none-any.whl", hash = "sha256:6437c62d4112b615ddebda362dfc16a4308d604c5b68125ed586e3e95d5b2e30", size = 494225, upload-time = "2026-02-17T14:05:18.596Z" },
+]
+
 [[package]]
 name = "setuptools"
 version = "80.10.2"
@@ -4852,6 +5197,15 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/9c/78/96ddb99933e11d91bc6e05edae23d2687e44213066bcbaca338898c73c47/textual-7.5.0-py3-none-any.whl", hash = "sha256:849dfee9d705eab3b2d07b33152b7bd74fb1f5056e002873cc448bce500c6374", size = 718164, upload-time = "2026-01-30T13:46:37.635Z" },
 ]
 
+[[package]]
+name = "threadpoolctl"
+version = "3.6.0"
+source = { registry = "https://pypi.org/simple" }
+sdist = { url = "https://files.pythonhosted.org/packages/b7/4d/08c89e34946fce2aec4fbb45c9016efd5f4d7f24af8e5d93296e935631d8/threadpoolctl-3.6.0.tar.gz", hash = "sha256:8ab8b4aa3491d812b623328249fab5302a68d2d71745c8a4c719a2fcaba9f44e", size = 21274, upload-time = "2025-03-13T13:49:23.031Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/32/d5/f9a850d79b0851d1d4ef6456097579a9005b31fea68726a4ae5f2d82ddd9/threadpoolctl-3.6.0-py3-none-any.whl", hash = "sha256:43a0b8fd5a2928500110039e43a5eed8480b918967083ea48dc3ab9f13c4a7fb", size = 18638, upload-time = "2025-03-13T13:49:21.846Z" },
+]
+
 [[package]]
 name = "tiktoken"
 version = "0.12.0"
@@ -4986,6 +5340,72 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/23/d1/136eb2cb77520a31e1f64cbae9d33ec6df0d78bdf4160398e86eec8a8754/tomli-2.4.0-py3-none-any.whl", hash = "sha256:1f776e7d669ebceb01dee46484485f43a4048746235e683bcdffacdf1fb4785a", size = 14477, upload-time = "2026-01-11T11:22:37.446Z" },
 ]
 
+[[package]]
+name = "torch"
+version = "2.10.0"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "cuda-bindings", marker = "python_full_version >= '3.14' and platform_machine == 'x86_64' and sys_platform == 'linux'" },
+    { name = "filelock", marker = "python_full_version >= '3.14'" },
+    { name = "fsspec", marker = "python_full_version >= '3.14'" },
+    { name = "jinja2", marker = "python_full_version >= '3.14'" },
+    { name = "networkx", marker = "python_full_version >= '3.14'" },
+    { name = "nvidia-cublas-cu12", marker = "python_full_version >= '3.14' and platform_machine == 'x86_64' and sys_platform == 'linux'" },
+    { name = "nvidia-cuda-cupti-cu12", marker = "python_full_version >= '3.14' and platform_machine == 'x86_64' and sys_platform == 'linux'" },
+    { name = "nvidia-cuda-nvrtc-cu12", marker = "python_full_version >= '3.14' and platform_machine == 'x86_64' and sys_platform == 'linux'" },
+    { name = "nvidia-cuda-runtime-cu12", marker = "python_full_version >= '3.14' and platform_machine == 'x86_64' and sys_platform == 'linux'" },
+    { name = "nvidia-cudnn-cu12", marker = "python_full_version >= '3.14' and platform_machine == 'x86_64' and sys_platform == 'linux'" },
+    { name = "nvidia-cufft-cu12", marker = "python_full_version >= '3.14' and platform_machine == 'x86_64' and sys_platform == 'linux'" },
+    { name = "nvidia-cufile-cu12", marker = "python_full_version >= '3.14' and platform_machine == 'x86_64' and sys_platform == 'linux'" },
+    { name = "nvidia-curand-cu12", marker = "python_full_version >= '3.14' and platform_machine == 'x86_64' and sys_platform == 'linux'" },
+    { name = "nvidia-cusolver-cu12", marker = "python_full_version >= '3.14' and platform_machine == 'x86_64' and sys_platform == 'linux'" },
+    { name = "nvidia-cusparse-cu12", marker = "python_full_version >= '3.14' and platform_machine == 'x86_64' and sys_platform == 'linux'" },
+    { name = "nvidia-cusparselt-cu12", marker = "python_full_version >= '3.14' and platform_machine == 'x86_64' and sys_platform == 'linux'" },
+    { name = "nvidia-nccl-cu12", marker = "python_full_version >= '3.14' and platform_machine == 'x86_64' and sys_platform == 'linux'" },
+    { name = "nvidia-nvjitlink-cu12", marker = "python_full_version >= '3.14' and platform_machine == 'x86_64' and sys_platform == 'linux'" },
+    { name = "nvidia-nvshmem-cu12", marker = "python_full_version >= '3.14' and platform_machine == 'x86_64' and sys_platform == 'linux'" },
+    { name = "nvidia-nvtx-cu12", marker = "python_full_version >= '3.14' and platform_machine == 'x86_64' and sys_platform == 'linux'" },
+    { name = "setuptools", marker = "python_full_version >= '3.14'" },
+    { name = "sympy", marker = "python_full_version >= '3.14'" },
+    { name = "triton", marker = "python_full_version >= '3.14' and platform_machine == 'x86_64' and sys_platform == 'linux'" },
+    { name = "typing-extensions", marker = "python_full_version >= '3.14'" },
+]
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/0f/8b/4b61d6e13f7108f36910df9ab4b58fd389cc2520d54d81b88660804aad99/torch-2.10.0-2-cp311-none-macosx_11_0_arm64.whl", hash = "sha256:418997cb02d0a0f1497cf6a09f63166f9f5df9f3e16c8a716ab76a72127c714f", size = 79423467, upload-time = "2026-02-10T21:44:48.711Z" },
+    { url = "https://files.pythonhosted.org/packages/d3/54/a2ba279afcca44bbd320d4e73675b282fcee3d81400ea1b53934efca6462/torch-2.10.0-2-cp312-none-macosx_11_0_arm64.whl", hash = "sha256:13ec4add8c3faaed8d13e0574f5cd4a323c11655546f91fbe6afa77b57423574", size = 79498202, upload-time = "2026-02-10T21:44:52.603Z" },
+    { url = "https://files.pythonhosted.org/packages/ec/23/2c9fe0c9c27f7f6cb865abcea8a4568f29f00acaeadfc6a37f6801f84cb4/torch-2.10.0-2-cp313-none-macosx_11_0_arm64.whl", hash = "sha256:e521c9f030a3774ed770a9c011751fb47c4d12029a3d6522116e48431f2ff89e", size = 79498254, upload-time = "2026-02-10T21:44:44.095Z" },
+    { url = "https://files.pythonhosted.org/packages/36/ab/7b562f1808d3f65414cd80a4f7d4bb00979d9355616c034c171249e1a303/torch-2.10.0-3-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:ac5bdcbb074384c66fa160c15b1ead77839e3fe7ed117d667249afce0acabfac", size = 915518691, upload-time = "2026-03-11T14:15:43.147Z" },
+    { url = "https://files.pythonhosted.org/packages/b3/7a/abada41517ce0011775f0f4eacc79659bc9bc6c361e6bfe6f7052a6b9363/torch-2.10.0-3-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:98c01b8bb5e3240426dcde1446eed6f40c778091c8544767ef1168fc663a05a6", size = 915622781, upload-time = "2026-03-11T14:17:11.354Z" },
+    { url = "https://files.pythonhosted.org/packages/ab/c6/4dfe238342ffdcec5aef1c96c457548762d33c40b45a1ab7033bb26d2ff2/torch-2.10.0-3-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:80b1b5bfe38eb0e9f5ff09f206dcac0a87aadd084230d4a36eea5ec5232c115b", size = 915627275, upload-time = "2026-03-11T14:16:11.325Z" },
+    { url = "https://files.pythonhosted.org/packages/d8/f0/72bf18847f58f877a6a8acf60614b14935e2f156d942483af1ffc081aea0/torch-2.10.0-3-cp313-cp313t-manylinux_2_28_x86_64.whl", hash = "sha256:46b3574d93a2a8134b3f5475cfb98e2eb46771794c57015f6ad1fb795ec25e49", size = 915523474, upload-time = "2026-03-11T14:17:44.422Z" },
+    { url = "https://files.pythonhosted.org/packages/f4/39/590742415c3030551944edc2ddc273ea1fdfe8ffb2780992e824f1ebee98/torch-2.10.0-3-cp314-cp314-manylinux_2_28_x86_64.whl", hash = "sha256:b1d5e2aba4eb7f8e87fbe04f86442887f9167a35f092afe4c237dfcaaef6e328", size = 915632474, upload-time = "2026-03-11T14:15:13.666Z" },
+    { url = "https://files.pythonhosted.org/packages/b6/8e/34949484f764dde5b222b7fe3fede43e4a6f0da9d7f8c370bb617d629ee2/torch-2.10.0-3-cp314-cp314t-manylinux_2_28_x86_64.whl", hash = "sha256:0228d20b06701c05a8f978357f657817a4a63984b0c90745def81c18aedfa591", size = 915523882, upload-time = "2026-03-11T14:14:46.311Z" },
+    { url = "https://files.pythonhosted.org/packages/78/89/f5554b13ebd71e05c0b002f95148033e730d3f7067f67423026cc9c69410/torch-2.10.0-cp311-cp311-manylinux_2_28_aarch64.whl", hash = "sha256:3282d9febd1e4e476630a099692b44fdc214ee9bf8ee5377732d9d9dfe5712e4", size = 145992610, upload-time = "2026-01-21T16:25:26.327Z" },
+    { url = "https://files.pythonhosted.org/packages/ae/30/a3a2120621bf9c17779b169fc17e3dc29b230c29d0f8222f499f5e159aa8/torch-2.10.0-cp311-cp311-manylinux_2_28_x86_64.whl", hash = "sha256:a2f9edd8dbc99f62bc4dfb78af7bf89499bca3d753423ac1b4e06592e467b763", size = 915607863, upload-time = "2026-01-21T16:25:06.696Z" },
+    { url = "https://files.pythonhosted.org/packages/6f/3d/c87b33c5f260a2a8ad68da7147e105f05868c281c63d65ed85aa4da98c66/torch-2.10.0-cp311-cp311-win_amd64.whl", hash = "sha256:29b7009dba4b7a1c960260fc8ac85022c784250af43af9fb0ebafc9883782ebd", size = 113723116, upload-time = "2026-01-21T16:25:21.916Z" },
+    { url = "https://files.pythonhosted.org/packages/61/d8/15b9d9d3a6b0c01b883787bd056acbe5cc321090d4b216d3ea89a8fcfdf3/torch-2.10.0-cp311-none-macosx_11_0_arm64.whl", hash = "sha256:b7bd80f3477b830dd166c707c5b0b82a898e7b16f59a7d9d42778dd058272e8b", size = 79423461, upload-time = "2026-01-21T16:24:50.266Z" },
+    { url = "https://files.pythonhosted.org/packages/cc/af/758e242e9102e9988969b5e621d41f36b8f258bb4a099109b7a4b4b50ea4/torch-2.10.0-cp312-cp312-manylinux_2_28_aarch64.whl", hash = "sha256:5fd4117d89ffd47e3dcc71e71a22efac24828ad781c7e46aaaf56bf7f2796acf", size = 145996088, upload-time = "2026-01-21T16:24:44.171Z" },
+    { url = "https://files.pythonhosted.org/packages/23/8e/3c74db5e53bff7ed9e34c8123e6a8bfef718b2450c35eefab85bb4a7e270/torch-2.10.0-cp312-cp312-manylinux_2_28_x86_64.whl", hash = "sha256:787124e7db3b379d4f1ed54dd12ae7c741c16a4d29b49c0226a89bea50923ffb", size = 915711952, upload-time = "2026-01-21T16:23:53.503Z" },
+    { url = "https://files.pythonhosted.org/packages/6e/01/624c4324ca01f66ae4c7cd1b74eb16fb52596dce66dbe51eff95ef9e7a4c/torch-2.10.0-cp312-cp312-win_amd64.whl", hash = "sha256:2c66c61f44c5f903046cc696d088e21062644cbe541c7f1c4eaae88b2ad23547", size = 113757972, upload-time = "2026-01-21T16:24:39.516Z" },
+    { url = "https://files.pythonhosted.org/packages/c9/5c/dee910b87c4d5c0fcb41b50839ae04df87c1cfc663cf1b5fca7ea565eeaa/torch-2.10.0-cp312-none-macosx_11_0_arm64.whl", hash = "sha256:6d3707a61863d1c4d6ebba7be4ca320f42b869ee657e9b2c21c736bf17000294", size = 79498198, upload-time = "2026-01-21T16:24:34.704Z" },
+    { url = "https://files.pythonhosted.org/packages/c9/6f/f2e91e34e3fcba2e3fc8d8f74e7d6c22e74e480bbd1db7bc8900fdf3e95c/torch-2.10.0-cp313-cp313-manylinux_2_28_aarch64.whl", hash = "sha256:5c4d217b14741e40776dd7074d9006fd28b8a97ef5654db959d8635b2fe5f29b", size = 146004247, upload-time = "2026-01-21T16:24:29.335Z" },
+    { url = "https://files.pythonhosted.org/packages/98/fb/5160261aeb5e1ee12ee95fe599d0541f7c976c3701d607d8fc29e623229f/torch-2.10.0-cp313-cp313-manylinux_2_28_x86_64.whl", hash = "sha256:6b71486353fce0f9714ca0c9ef1c850a2ae766b409808acd58e9678a3edb7738", size = 915716445, upload-time = "2026-01-21T16:22:45.353Z" },
+    { url = "https://files.pythonhosted.org/packages/6a/16/502fb1b41e6d868e8deb5b0e3ae926bbb36dab8ceb0d1b769b266ad7b0c3/torch-2.10.0-cp313-cp313-win_amd64.whl", hash = "sha256:c2ee399c644dc92ef7bc0d4f7e74b5360c37cdbe7c5ba11318dda49ffac2bc57", size = 113757050, upload-time = "2026-01-21T16:24:19.204Z" },
+    { url = "https://files.pythonhosted.org/packages/1a/0b/39929b148f4824bc3ad6f9f72a29d4ad865bcf7ebfc2fa67584773e083d2/torch-2.10.0-cp313-cp313t-macosx_14_0_arm64.whl", hash = "sha256:3202429f58309b9fa96a614885eace4b7995729f44beb54d3e4a47773649d382", size = 79851305, upload-time = "2026-01-21T16:24:09.209Z" },
+    { url = "https://files.pythonhosted.org/packages/d8/14/21fbce63bc452381ba5f74a2c0a959fdf5ad5803ccc0c654e752e0dbe91a/torch-2.10.0-cp313-cp313t-manylinux_2_28_aarch64.whl", hash = "sha256:aae1b29cd68e50a9397f5ee897b9c24742e9e306f88a807a27d617f07adb3bd8", size = 146005472, upload-time = "2026-01-21T16:22:29.022Z" },
+    { url = "https://files.pythonhosted.org/packages/54/fd/b207d1c525cb570ef47f3e9f836b154685011fce11a2f444ba8a4084d042/torch-2.10.0-cp313-cp313t-manylinux_2_28_x86_64.whl", hash = "sha256:6021db85958db2f07ec94e1bc77212721ba4920c12a18dc552d2ae36a3eb163f", size = 915612644, upload-time = "2026-01-21T16:21:47.019Z" },
+    { url = "https://files.pythonhosted.org/packages/36/53/0197f868c75f1050b199fe58f9bf3bf3aecac9b4e85cc9c964383d745403/torch-2.10.0-cp313-cp313t-win_amd64.whl", hash = "sha256:ff43db38af76fda183156153983c9a096fc4c78d0cd1e07b14a2314c7f01c2c8", size = 113997015, upload-time = "2026-01-21T16:23:00.767Z" },
+    { url = "https://files.pythonhosted.org/packages/0e/13/e76b4d9c160e89fff48bf16b449ea324bda84745d2ab30294c37c2434c0d/torch-2.10.0-cp313-none-macosx_11_0_arm64.whl", hash = "sha256:cdf2a523d699b70d613243211ecaac14fe9c5df8a0b0a9c02add60fb2a413e0f", size = 79498248, upload-time = "2026-01-21T16:23:09.315Z" },
+    { url = "https://files.pythonhosted.org/packages/4f/93/716b5ac0155f1be70ed81bacc21269c3ece8dba0c249b9994094110bfc51/torch-2.10.0-cp314-cp314-macosx_14_0_arm64.whl", hash = "sha256:bf0d9ff448b0218e0433aeb198805192346c4fd659c852370d5cc245f602a06a", size = 79464992, upload-time = "2026-01-21T16:23:05.162Z" },
+    { url = "https://files.pythonhosted.org/packages/69/2b/51e663ff190c9d16d4a8271203b71bc73a16aa7619b9f271a69b9d4a936b/torch-2.10.0-cp314-cp314-manylinux_2_28_aarch64.whl", hash = "sha256:233aed0659a2503b831d8a67e9da66a62c996204c0bba4f4c442ccc0c68a3f60", size = 146018567, upload-time = "2026-01-21T16:22:23.393Z" },
+    { url = "https://files.pythonhosted.org/packages/5e/cd/4b95ef7f293b927c283db0b136c42be91c8ec6845c44de0238c8c23bdc80/torch-2.10.0-cp314-cp314-manylinux_2_28_x86_64.whl", hash = "sha256:682497e16bdfa6efeec8cde66531bc8d1fbbbb4d8788ec6173c089ed3cc2bfe5", size = 915721646, upload-time = "2026-01-21T16:21:16.983Z" },
+    { url = "https://files.pythonhosted.org/packages/56/97/078a007208f8056d88ae43198833469e61a0a355abc0b070edd2c085eb9a/torch-2.10.0-cp314-cp314-win_amd64.whl", hash = "sha256:6528f13d2a8593a1a412ea07a99812495bec07e9224c28b2a25c0a30c7da025c", size = 113752373, upload-time = "2026-01-21T16:22:13.471Z" },
+    { url = "https://files.pythonhosted.org/packages/d8/94/71994e7d0d5238393df9732fdab607e37e2b56d26a746cb59fdb415f8966/torch-2.10.0-cp314-cp314t-macosx_14_0_arm64.whl", hash = "sha256:f5ab4ba32383061be0fb74bda772d470140a12c1c3b58a0cfbf3dae94d164c28", size = 79850324, upload-time = "2026-01-21T16:22:09.494Z" },
+    { url = "https://files.pythonhosted.org/packages/e2/65/1a05346b418ea8ccd10360eef4b3e0ce688fba544e76edec26913a8d0ee0/torch-2.10.0-cp314-cp314t-manylinux_2_28_aarch64.whl", hash = "sha256:716b01a176c2a5659c98f6b01bf868244abdd896526f1c692712ab36dbaf9b63", size = 146006482, upload-time = "2026-01-21T16:22:18.42Z" },
+    { url = "https://files.pythonhosted.org/packages/1d/b9/5f6f9d9e859fc3235f60578fa64f52c9c6e9b4327f0fe0defb6de5c0de31/torch-2.10.0-cp314-cp314t-manylinux_2_28_x86_64.whl", hash = "sha256:d8f5912ba938233f86361e891789595ff35ca4b4e2ac8fe3670895e5976731d6", size = 915613050, upload-time = "2026-01-21T16:20:49.035Z" },
+    { url = "https://files.pythonhosted.org/packages/66/4d/35352043ee0eaffdeff154fad67cd4a31dbed7ff8e3be1cc4549717d6d51/torch-2.10.0-cp314-cp314t-win_amd64.whl", hash = "sha256:71283a373f0ee2c89e0f0d5f446039bdabe8dbc3c9ccf35f0f784908b0acd185", size = 113995816, upload-time = "2026-01-21T16:22:05.312Z" },
+]
+
 [[package]]
 name = "tqdm"
 version = "4.67.2"
@@ -4998,6 +5418,39 @@ wheels = [
     { url = "https://files.pythonhosted.org/packages/f5/e2/31eac96de2915cf20ccaed0225035db149dfb9165a9ed28d4b252ef3f7f7/tqdm-4.67.2-py3-none-any.whl", hash = "sha256:9a12abcbbff58b6036b2167d9d3853042b9d436fe7330f06ae047867f2f8e0a7", size = 78354, upload-time = "2026-01-30T23:12:04.368Z" },
 ]
 
+[[package]]
+name = "transformers"
+version = "5.3.0"
+source = { registry = "https://pypi.org/simple" }
+dependencies = [
+    { name = "huggingface-hub", marker = "python_full_version >= '3.14'" },
+    { name = "numpy", marker = "python_full_version >= '3.14'" },
+    { name = "packaging", marker = "python_full_version >= '3.14'" },
+    { name = "pyyaml", marker = "python_full_version >= '3.14'" },
+    { name = "regex", marker = "python_full_version >= '3.14'" },
+    { name = "safetensors", marker = "python_full_version >= '3.14'" },
+    { name = "tokenizers", marker = "python_full_version >= '3.14'" },
+    { name = "tqdm", marker = "python_full_version >= '3.14'" },
+    { name = "typer", marker = "python_full_version >= '3.14'" },
+]
+sdist = { url = "https://files.pythonhosted.org/packages/fc/1a/70e830d53ecc96ce69cfa8de38f163712d2b43ac52fbd743f39f56025c31/transformers-5.3.0.tar.gz", hash = "sha256:009555b364029da9e2946d41f1c5de9f15e6b1df46b189b7293f33a161b9c557", size = 8830831, upload-time = "2026-03-04T17:41:46.119Z" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/b8/88/ae8320064e32679a5429a2c9ebbc05c2bf32cefb6e076f9b07f6d685a9b4/transformers-5.3.0-py3-none-any.whl", hash = "sha256:50ac8c89c3c7033444fb3f9f53138096b997ebb70d4b5e50a2e810bf12d3d29a", size = 10661827, upload-time = "2026-03-04T17:41:42.722Z" },
+]
+
+[[package]]
+name = "triton"
+version = "3.6.0"
+source = { registry = "https://pypi.org/simple" }
+wheels = [
+    { url = "https://files.pythonhosted.org/packages/e0/12/b05ba554d2c623bffa59922b94b0775673de251f468a9609bc9e45de95e9/triton-3.6.0-cp311-cp311-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:e8e323d608e3a9bfcc2d9efcc90ceefb764a82b99dea12a86d643c72539ad5d3", size = 188214640, upload-time = "2026-01-20T16:00:35.869Z" },
+    { url = "https://files.pythonhosted.org/packages/ab/a8/cdf8b3e4c98132f965f88c2313a4b493266832ad47fb52f23d14d4f86bb5/triton-3.6.0-cp312-cp312-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:74caf5e34b66d9f3a429af689c1c7128daba1d8208df60e81106b115c00d6fca", size = 188266850, upload-time = "2026-01-20T16:00:43.041Z" },
+    { url = "https://files.pythonhosted.org/packages/f9/0b/37d991d8c130ce81a8728ae3c25b6e60935838e9be1b58791f5997b24a54/triton-3.6.0-cp313-cp313-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:10c7f76c6e72d2ef08df639e3d0d30729112f47a56b0c81672edc05ee5116ac9", size = 188289450, upload-time = "2026-01-20T16:00:49.136Z" },
+    { url = "https://files.pythonhosted.org/packages/35/f8/9c66bfc55361ec6d0e4040a0337fb5924ceb23de4648b8a81ae9d33b2b38/triton-3.6.0-cp313-cp313t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:d002e07d7180fd65e622134fbd980c9a3d4211fb85224b56a0a0efbd422ab72f", size = 188400296, upload-time = "2026-01-20T16:00:56.042Z" },
+    { url = "https://files.pythonhosted.org/packages/df/3d/9e7eee57b37c80cec63322c0231bb6da3cfe535a91d7a4d64896fcb89357/triton-3.6.0-cp314-cp314-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:a17a5d5985f0ac494ed8a8e54568f092f7057ef60e1b0fa09d3fd1512064e803", size = 188273063, upload-time = "2026-01-20T16:01:07.278Z" },
+    { url = "https://files.pythonhosted.org/packages/f6/56/6113c23ff46c00aae423333eb58b3e60bdfe9179d542781955a5e1514cb3/triton-3.6.0-cp314-cp314t-manylinux_2_27_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:46bd1c1af4b6704e554cad2eeb3b0a6513a980d470ccfa63189737340c7746a7", size = 188397994, upload-time = "2026-01-20T16:01:14.236Z" },
+]
+
 [[package]]
 name = "typer"
 version = "0.21.1"
@@ -5419,14 +5872,14 @@ wheels = [
 
 [[package]]
 name = "werkzeug"
-version = "3.1.5"
+version = "3.1.6"
 source = { registry = "https://pypi.org/simple" }
 dependencies = [
     { name = "markupsafe" },
 ]
-sdist = { url = "https://files.pythonhosted.org/packages/5a/70/1469ef1d3542ae7c2c7b72bd5e3a4e6ee69d7978fa8a3af05a38eca5becf/werkzeug-3.1.5.tar.gz", hash = "sha256:6a548b0e88955dd07ccb25539d7d0cc97417ee9e179677d22c7041c8f078ce67", size = 864754, upload-time = "2026-01-08T17:49:23.247Z" }
+sdist = { url = "https://files.pythonhosted.org/packages/61/f1/ee81806690a87dab5f5653c1f146c92bc066d7f4cebc603ef88eb9e13957/werkzeug-3.1.6.tar.gz", hash = "sha256:210c6bede5a420a913956b4791a7f4d6843a43b6fcee4dfa08a65e93007d0d25", size = 864736, upload-time = "2026-02-19T15:17:18.884Z" }
 wheels = [
-    { url = "https://files.pythonhosted.org/packages/ad/e4/8d97cca767bcc1be76d16fb76951608305561c6e056811587f36cb1316a8/werkzeug-3.1.5-py3-none-any.whl", hash = "sha256:5111e36e91086ece91f93268bb39b4a35c1e6f1feac762c9c822ded0a4e322dc", size = 225025, upload-time = "2026-01-08T17:49:21.859Z" },
+    { url = "https://files.pythonhosted.org/packages/4d/ec/d58832f89ede95652fd01f4f24236af7d32b70cab2196dfcc2d2fd13c5c2/werkzeug-3.1.6-py3-none-any.whl", hash = "sha256:7ddf3357bb9564e407607f988f683d72038551200c704012bb9a4c523d42f131", size = 225166, upload-time = "2026-02-19T15:17:17.475Z" },
 ]
 
 [[package]]

web/package-lock.json

@@ -32,7 +32,7 @@
         "@radix-ui/react-tooltip": "^1.2.7",
         "@tailwindcss/postcss": "^4.1.5",
         "@tanstack/react-table": "^8.21.3",
-        "axios": "^1.12.0",
+        "axios": "^1.13.5",
         "class-variance-authority": "^0.7.1",
         "clsx": "^2.1.1",
         "highlight.js": "^11.11.1",
@@ -56,6 +56,7 @@
         "rehype-autolink-headings": "^7.1.0",
         "rehype-highlight": "^7.0.2",
         "rehype-raw": "^7.0.0",
+        "rehype-sanitize": "^6.0.0",
         "rehype-slug": "^6.0.0",
         "remark-gfm": "^4.0.1",
         "sonner": "^2.0.3",
@@ -3798,13 +3799,13 @@
       }
     },
     "node_modules/axios": {
-      "version": "1.13.4",
-      "resolved": "https://registry.npmjs.org/axios/-/axios-1.13.4.tgz",
-      "integrity": "sha512-1wVkUaAO6WyaYtCkcYCOx12ZgpGf9Zif+qXa4n+oYzK558YryKqiL6UWwd5DqiH3VRW0GYhTZQ/vlgJrCoNQlg==",
+      "version": "1.13.6",
+      "resolved": "https://registry.npmjs.org/axios/-/axios-1.13.6.tgz",
+      "integrity": "sha512-ChTCHMouEe2kn713WHbQGcuYrr6fXTBiu460OTwWrWob16g1bXn4vtz07Ope7ewMozJAnEquLk5lWQWtBig9DQ==",
       "license": "MIT",
       "dependencies": {
-        "follow-redirects": "^1.15.6",
-        "form-data": "^4.0.4",
+        "follow-redirects": "^1.15.11",
+        "form-data": "^4.0.5",
         "proxy-from-env": "^1.1.0"
       }
     },
@@ -5970,6 +5971,21 @@
         "url": "https://opencollective.com/unified"
       }
     },
+    "node_modules/hast-util-sanitize": {
+      "version": "5.0.2",
+      "resolved": "https://registry.npmjs.org/hast-util-sanitize/-/hast-util-sanitize-5.0.2.tgz",
+      "integrity": "sha512-3yTWghByc50aGS7JlGhk61SPenfE/p1oaFeNwkOOyrscaOkMGrcW9+Cy/QAIOBpZxP1yqDIzFMR0+Np0i0+usg==",
+      "license": "MIT",
+      "dependencies": {
+        "@types/hast": "^3.0.0",
+        "@ungap/structured-clone": "^1.0.0",
+        "unist-util-position": "^5.0.0"
+      },
+      "funding": {
+        "type": "opencollective",
+        "url": "https://opencollective.com/unified"
+      }
+    },
     "node_modules/hast-util-to-jsx-runtime": {
       "version": "2.3.6",
       "resolved": "https://registry.npmjs.org/hast-util-to-jsx-runtime/-/hast-util-to-jsx-runtime-2.3.6.tgz",
@@ -9392,6 +9408,20 @@
         "url": "https://opencollective.com/unified"
       }
     },
+    "node_modules/rehype-sanitize": {
+      "version": "6.0.0",
+      "resolved": "https://registry.npmjs.org/rehype-sanitize/-/rehype-sanitize-6.0.0.tgz",
+      "integrity": "sha512-CsnhKNsyI8Tub6L4sm5ZFsme4puGfc6pYylvXo1AeqaGbjOYyzNv3qZPwvs0oMJ39eryyeOdmxwUIo94IpEhqg==",
+      "license": "MIT",
+      "dependencies": {
+        "@types/hast": "^3.0.0",
+        "hast-util-sanitize": "^5.0.0"
+      },
+      "funding": {
+        "type": "opencollective",
+        "url": "https://opencollective.com/unified"
+      }
+    },
     "node_modules/rehype-slug": {
       "version": "6.0.0",
       "resolved": "https://registry.npmjs.org/rehype-slug/-/rehype-slug-6.0.0.tgz",

web/package.json

@@ -6,8 +6,8 @@
     "dev": "next dev --turbopack",
     "build": "next build",
     "start": "next start",
-    "lint": "eslint .",
-    "lint:fix": "eslint . --fix",
+    "lint": "eslint src",
+    "lint:fix": "eslint src --fix",
     "lint-staged": "lint-staged"
   },
   "lint-staged": {
@@ -68,6 +68,7 @@
     "rehype-autolink-headings": "^7.1.0",
     "rehype-highlight": "^7.0.2",
     "rehype-raw": "^7.0.0",
+    "rehype-sanitize": "^6.0.0",
     "rehype-slug": "^6.0.0",
     "remark-gfm": "^4.0.1",
     "sonner": "^2.0.3",
@@ -101,5 +102,10 @@
     "typescript": "^5.8.3",
     "typescript-eslint": "^8.31.1"
   },
-  "packageManager": "pnpm@8.9.2+sha512.b9d35fe91b2a5854dadc43034a3e7b2e675fa4b56e20e8e09ef078fa553c18f8aed44051e7b36e8b8dd435f97eb0c44c4ff3b44fc7c6fa7d21e1fac17bbe661e"
+  "packageManager": "pnpm@8.9.2+sha512.b9d35fe91b2a5854dadc43034a3e7b2e675fa4b56e20e8e09ef078fa553c18f8aed44051e7b36e8b8dd435f97eb0c44c4ff3b44fc7c6fa7d21e1fac17bbe661e",
+  "pnpm": {
+    "overrides": {
+      "minimatch": "3.1.3"
+    }
+  }
 }

web/pnpm-lock.yaml

@@ -4,6 +4,9 @@ settings:
   autoInstallPeers: true
   excludeLinksFromLockfile: false
 
+overrides:
+  minimatch: 3.1.3
+
 dependencies:
   '@dnd-kit/core':
     specifier: ^6.3.1
@@ -149,6 +152,9 @@ dependencies:
   rehype-raw:
     specifier: ^7.0.0
     version: 7.0.0
+  rehype-sanitize:
+    specifier: ^6.0.0
+    version: 6.0.0
   rehype-slug:
     specifier: ^6.0.0
     version: 6.0.0
@@ -342,7 +348,7 @@ packages:
     dependencies:
       '@eslint/object-schema': 2.1.7
       debug: 4.4.3
-      minimatch: 3.1.2
+      minimatch: 3.1.3
     transitivePeerDependencies:
       - supports-color
     dev: true
@@ -372,7 +378,7 @@ packages:
       ignore: 5.3.2
       import-fresh: 3.3.1
       js-yaml: 4.1.1
-      minimatch: 3.1.2
+      minimatch: 3.1.3
       strip-json-comments: 3.1.1
     transitivePeerDependencies:
       - supports-color
@@ -505,6 +511,7 @@ packages:
     resolution: {integrity: sha512-excjX8DfsIcJ10x1Kzr4RcWe1edC9PquDRRPx3YVCvQv+U5p7Yin2s32ftzikXojb1PIFc/9Mt28/y+iRklkrw==}
     cpu: [arm64]
     os: [linux]
+    libc: [glibc]
     requiresBuild: true
     dev: false
     optional: true
@@ -513,6 +520,7 @@ packages:
     resolution: {integrity: sha512-bFI7xcKFELdiNCVov8e44Ia4u2byA+l3XtsAj+Q8tfCwO6BQ8iDojYdvoPMqsKDkuoOo+X6HZA0s0q11ANMQ8A==}
     cpu: [arm]
     os: [linux]
+    libc: [glibc]
     requiresBuild: true
     dev: false
     optional: true
@@ -521,6 +529,7 @@ packages:
     resolution: {integrity: sha512-FMuvGijLDYG6lW+b/UvyilUWu5Ayu+3r2d1S8notiGCIyYU/76eig1UfMmkZ7vwgOrzKzlQbFSuQfgm7GYUPpA==}
     cpu: [ppc64]
     os: [linux]
+    libc: [glibc]
     requiresBuild: true
     dev: false
     optional: true
@@ -529,6 +538,7 @@ packages:
     resolution: {integrity: sha512-oVDbcR4zUC0ce82teubSm+x6ETixtKZBh/qbREIOcI3cULzDyb18Sr/Wcyx7NRQeQzOiHTNbZFF1UwPS2scyGA==}
     cpu: [riscv64]
     os: [linux]
+    libc: [glibc]
     requiresBuild: true
     dev: false
     optional: true
@@ -537,6 +547,7 @@ packages:
     resolution: {integrity: sha512-qmp9VrzgPgMoGZyPvrQHqk02uyjA0/QrTO26Tqk6l4ZV0MPWIW6LTkqOIov+J1yEu7MbFQaDpwdwJKhbJvuRxQ==}
     cpu: [s390x]
     os: [linux]
+    libc: [glibc]
     requiresBuild: true
     dev: false
     optional: true
@@ -545,6 +556,7 @@ packages:
     resolution: {integrity: sha512-tJxiiLsmHc9Ax1bz3oaOYBURTXGIRDODBqhveVHonrHJ9/+k89qbLl0bcJns+e4t4rvaNBxaEZsFtSfAdquPrw==}
     cpu: [x64]
     os: [linux]
+    libc: [glibc]
     requiresBuild: true
     dev: false
     optional: true
@@ -553,6 +565,7 @@ packages:
     resolution: {integrity: sha512-FVQHuwx1IIuNow9QAbYUzJ+En8KcVm9Lk5+uGUQJHaZmMECZmOlix9HnH7n1TRkXMS0pGxIJokIVB9SuqZGGXw==}
     cpu: [arm64]
     os: [linux]
+    libc: [musl]
     requiresBuild: true
     dev: false
     optional: true
@@ -561,6 +574,7 @@ packages:
     resolution: {integrity: sha512-+LpyBk7L44ZIXwz/VYfglaX/okxezESc6UxDSoyo2Ks6Jxc4Y7sGjpgU9s4PMgqgjj1gZCylTieNamqA1MF7Dg==}
     cpu: [x64]
     os: [linux]
+    libc: [musl]
     requiresBuild: true
     dev: false
     optional: true
@@ -570,6 +584,7 @@ packages:
     engines: {node: ^18.17.0 || ^20.3.0 || >=21.0.0}
     cpu: [arm64]
     os: [linux]
+    libc: [glibc]
     requiresBuild: true
     optionalDependencies:
       '@img/sharp-libvips-linux-arm64': 1.2.4
@@ -581,6 +596,7 @@ packages:
     engines: {node: ^18.17.0 || ^20.3.0 || >=21.0.0}
     cpu: [arm]
     os: [linux]
+    libc: [glibc]
     requiresBuild: true
     optionalDependencies:
       '@img/sharp-libvips-linux-arm': 1.2.4
@@ -592,6 +608,7 @@ packages:
     engines: {node: ^18.17.0 || ^20.3.0 || >=21.0.0}
     cpu: [ppc64]
     os: [linux]
+    libc: [glibc]
     requiresBuild: true
     optionalDependencies:
       '@img/sharp-libvips-linux-ppc64': 1.2.4
@@ -603,6 +620,7 @@ packages:
     engines: {node: ^18.17.0 || ^20.3.0 || >=21.0.0}
     cpu: [riscv64]
     os: [linux]
+    libc: [glibc]
     requiresBuild: true
     optionalDependencies:
       '@img/sharp-libvips-linux-riscv64': 1.2.4
@@ -614,6 +632,7 @@ packages:
     engines: {node: ^18.17.0 || ^20.3.0 || >=21.0.0}
     cpu: [s390x]
     os: [linux]
+    libc: [glibc]
     requiresBuild: true
     optionalDependencies:
       '@img/sharp-libvips-linux-s390x': 1.2.4
@@ -625,6 +644,7 @@ packages:
     engines: {node: ^18.17.0 || ^20.3.0 || >=21.0.0}
     cpu: [x64]
     os: [linux]
+    libc: [glibc]
     requiresBuild: true
     optionalDependencies:
       '@img/sharp-libvips-linux-x64': 1.2.4
@@ -636,6 +656,7 @@ packages:
     engines: {node: ^18.17.0 || ^20.3.0 || >=21.0.0}
     cpu: [arm64]
     os: [linux]
+    libc: [musl]
     requiresBuild: true
     optionalDependencies:
       '@img/sharp-libvips-linuxmusl-arm64': 1.2.4
@@ -647,6 +668,7 @@ packages:
     engines: {node: ^18.17.0 || ^20.3.0 || >=21.0.0}
     cpu: [x64]
     os: [linux]
+    libc: [musl]
     requiresBuild: true
     optionalDependencies:
       '@img/sharp-libvips-linuxmusl-x64': 1.2.4
@@ -763,6 +785,7 @@ packages:
     engines: {node: '>= 10'}
     cpu: [arm64]
     os: [linux]
+    libc: [glibc]
     requiresBuild: true
     dev: false
     optional: true
@@ -772,6 +795,7 @@ packages:
     engines: {node: '>= 10'}
     cpu: [arm64]
     os: [linux]
+    libc: [musl]
     requiresBuild: true
     dev: false
     optional: true
@@ -781,6 +805,7 @@ packages:
     engines: {node: '>= 10'}
     cpu: [x64]
     os: [linux]
+    libc: [glibc]
     requiresBuild: true
     dev: false
     optional: true
@@ -790,6 +815,7 @@ packages:
     engines: {node: '>= 10'}
     cpu: [x64]
     os: [linux]
+    libc: [musl]
     requiresBuild: true
     dev: false
     optional: true
@@ -1889,6 +1915,7 @@ packages:
     engines: {node: '>= 10'}
     cpu: [arm64]
     os: [linux]
+    libc: [glibc]
     requiresBuild: true
     dev: false
     optional: true
@@ -1898,6 +1925,7 @@ packages:
     engines: {node: '>= 10'}
     cpu: [arm64]
     os: [linux]
+    libc: [musl]
     requiresBuild: true
     dev: false
     optional: true
@@ -1907,6 +1935,7 @@ packages:
     engines: {node: '>= 10'}
     cpu: [x64]
     os: [linux]
+    libc: [glibc]
     requiresBuild: true
     dev: false
     optional: true
@@ -1916,6 +1945,7 @@ packages:
     engines: {node: '>= 10'}
     cpu: [x64]
     os: [linux]
+    libc: [musl]
     requiresBuild: true
     dev: false
     optional: true
@@ -2233,7 +2263,7 @@ packages:
       '@typescript-eslint/types': 8.54.0
       '@typescript-eslint/visitor-keys': 8.54.0
       debug: 4.4.3
-      minimatch: 9.0.5
+      minimatch: 3.1.3
       semver: 7.7.3
       tinyglobby: 0.2.15
       ts-api-utils: 2.4.0(typescript@5.9.3)
@@ -2331,6 +2361,7 @@ packages:
     resolution: {integrity: sha512-34gw7PjDGB9JgePJEmhEqBhWvCiiWCuXsL9hYphDF7crW7UgI05gyBAi6MF58uGcMOiOqSJ2ybEeCvHcq0BCmQ==}
     cpu: [arm64]
     os: [linux]
+    libc: [glibc]
     requiresBuild: true
     dev: true
     optional: true
@@ -2339,6 +2370,7 @@ packages:
     resolution: {integrity: sha512-RyMIx6Uf53hhOtJDIamSbTskA99sPHS96wxVE/bJtePJJtpdKGXO1wY90oRdXuYOGOTuqjT8ACccMc4K6QmT3w==}
     cpu: [arm64]
     os: [linux]
+    libc: [musl]
     requiresBuild: true
     dev: true
     optional: true
@@ -2347,6 +2379,7 @@ packages:
     resolution: {integrity: sha512-D8Vae74A4/a+mZH0FbOkFJL9DSK2R6TFPC9M+jCWYia/q2einCubX10pecpDiTmkJVUH+y8K3BZClycD8nCShA==}
     cpu: [ppc64]
     os: [linux]
+    libc: [glibc]
     requiresBuild: true
     dev: true
     optional: true
@@ -2355,6 +2388,7 @@ packages:
     resolution: {integrity: sha512-frxL4OrzOWVVsOc96+V3aqTIQl1O2TjgExV4EKgRY09AJ9leZpEg8Ak9phadbuX0BA4k8U5qtvMSQQGGmaJqcQ==}
     cpu: [riscv64]
     os: [linux]
+    libc: [glibc]
     requiresBuild: true
     dev: true
     optional: true
@@ -2363,6 +2397,7 @@ packages:
     resolution: {integrity: sha512-mJ5vuDaIZ+l/acv01sHoXfpnyrNKOk/3aDoEdLO/Xtn9HuZlDD6jKxHlkN8ZhWyLJsRBxfv9GYM2utQ1SChKew==}
     cpu: [riscv64]
     os: [linux]
+    libc: [musl]
     requiresBuild: true
     dev: true
     optional: true
@@ -2371,6 +2406,7 @@ packages:
     resolution: {integrity: sha512-kELo8ebBVtb9sA7rMe1Cph4QHreByhaZ2QEADd9NzIQsYNQpt9UkM9iqr2lhGr5afh885d/cB5QeTXSbZHTYPg==}
     cpu: [s390x]
     os: [linux]
+    libc: [glibc]
     requiresBuild: true
     dev: true
     optional: true
@@ -2379,6 +2415,7 @@ packages:
     resolution: {integrity: sha512-C3ZAHugKgovV5YvAMsxhq0gtXuwESUKc5MhEtjBpLoHPLYM+iuwSj3lflFwK3DPm68660rZ7G8BMcwSro7hD5w==}
     cpu: [x64]
     os: [linux]
+    libc: [glibc]
     requiresBuild: true
     dev: true
     optional: true
@@ -2387,6 +2424,7 @@ packages:
     resolution: {integrity: sha512-rV0YSoyhK2nZ4vEswT/QwqzqQXw5I6CjoaYMOX0TqBlWhojUf8P94mvI7nuJTeaCkkds3QE4+zS8Ko+GdXuZtA==}
     cpu: [x64]
     os: [linux]
+    libc: [musl]
     requiresBuild: true
     dev: true
     optional: true
@@ -2643,12 +2681,6 @@ packages:
       concat-map: 0.0.1
     dev: true
 
-  /brace-expansion@2.0.2:
-    resolution: {integrity: sha512-Jt0vHyM+jmUBqojB7E1NIYadt0vI0Qxjxd2TErW94wDz+E2LAm5vKMXXwg6ZZBTHPuUlDgQHKXvjGBdfcF1ZDQ==}
-    dependencies:
-      balanced-match: 1.0.2
-    dev: true
-
   /braces@3.0.3:
     resolution: {integrity: sha512-yQbXgO/OSZVD2IsiLlro+7Hf6Q18EJrKSEsdoMzKePKXct3gvD8oLcOQdIzGupr5Fj+EDe8gO/lxc1BzfMpxvA==}
     engines: {node: '>=8'}
@@ -3310,7 +3342,7 @@ packages:
       hasown: 2.0.2
       is-core-module: 2.16.1
       is-glob: 4.0.3
-      minimatch: 3.1.2
+      minimatch: 3.1.3
       object.fromentries: 2.0.8
       object.groupby: 1.0.3
       object.values: 1.2.1
@@ -3341,7 +3373,7 @@ packages:
       hasown: 2.0.2
       jsx-ast-utils: 3.3.5
       language-tags: 1.0.9
-      minimatch: 3.1.2
+      minimatch: 3.1.3
       object.fromentries: 2.0.8
       safe-regex-test: 1.1.0
       string.prototype.includes: 2.0.1
@@ -3393,7 +3425,7 @@ packages:
       estraverse: 5.3.0
       hasown: 2.0.2
       jsx-ast-utils: 3.3.5
-      minimatch: 3.1.2
+      minimatch: 3.1.3
       object.entries: 1.1.9
       object.fromentries: 2.0.8
       object.values: 1.2.1
@@ -3463,7 +3495,7 @@ packages:
       is-glob: 4.0.3
       json-stable-stringify-without-jsonify: 1.0.1
       lodash.merge: 4.6.2
-      minimatch: 3.1.2
+      minimatch: 3.1.3
       natural-compare: 1.4.0
       optionator: 0.9.4
     transitivePeerDependencies:
@@ -3873,6 +3905,14 @@ packages:
       zwitch: 2.0.4
     dev: false
 
+  /hast-util-sanitize@5.0.2:
+    resolution: {integrity: sha512-3yTWghByc50aGS7JlGhk61SPenfE/p1oaFeNwkOOyrscaOkMGrcW9+Cy/QAIOBpZxP1yqDIzFMR0+Np0i0+usg==}
+    dependencies:
+      '@types/hast': 3.0.4
+      '@ungap/structured-clone': 1.3.0
+      unist-util-position: 5.0.0
+    dev: false
+
   /hast-util-to-jsx-runtime@2.3.6:
     resolution: {integrity: sha512-zl6s8LwNyo1P9uw+XJGvZtdFF1GdAkOg8ujOw+4Pyb76874fLps4ueHXDhXWdk6YHQ6OgUtinliG7RsYvCbbBg==}
     dependencies:
@@ -4413,6 +4453,7 @@ packages:
     engines: {node: '>= 12.0.0'}
     cpu: [arm64]
     os: [linux]
+    libc: [glibc]
     requiresBuild: true
     dev: false
     optional: true
@@ -4422,6 +4463,7 @@ packages:
     engines: {node: '>= 12.0.0'}
     cpu: [arm64]
     os: [linux]
+    libc: [musl]
     requiresBuild: true
     dev: false
     optional: true
@@ -4431,6 +4473,7 @@ packages:
     engines: {node: '>= 12.0.0'}
     cpu: [x64]
     os: [linux]
+    libc: [glibc]
     requiresBuild: true
     dev: false
     optional: true
@@ -4440,6 +4483,7 @@ packages:
     engines: {node: '>= 12.0.0'}
     cpu: [x64]
     os: [linux]
+    libc: [musl]
     requiresBuild: true
     dev: false
     optional: true
@@ -5066,19 +5110,12 @@ packages:
     engines: {node: '>=18'}
     dev: true
 
-  /minimatch@3.1.2:
-    resolution: {integrity: sha512-J7p63hRiAjw1NDEww1W7i37+ByIrOWO5XQQAzZ3VOcL0PNybwpfmV/N05zFAzwQ9USyEcX6t3UO+K5aqBQOIHw==}
+  /minimatch@3.1.3:
+    resolution: {integrity: sha512-M2GCs7Vk83NxkUyQV1bkABc4yxgz9kILhHImZiBPAZ9ybuvCb0/H7lEl5XvIg3g+9d4eNotkZA5IWwYl0tibaA==}
     dependencies:
       brace-expansion: 1.1.12
     dev: true
 
-  /minimatch@9.0.5:
-    resolution: {integrity: sha512-G6T0ZX48xgozx7587koeX9Ys2NYy6Gmv//P89sEte9V9whIapMNF4idKxnW2QtCcLiTWlb/wfCabAtAFWhhBow==}
-    engines: {node: '>=16 || 14 >=14.17'}
-    dependencies:
-      brace-expansion: 2.0.2
-    dev: true
-
   /minimist@1.2.8:
     resolution: {integrity: sha512-2yyAR8qBkN3YuheJanUpWC5U3bb5osDywNB8RzDVlDwDHbocAJveqqj1u8+SVD7jkWT4yvsHCpWqqWqAxb0zCA==}
     dev: true
@@ -5713,6 +5750,13 @@ packages:
       vfile: 6.0.3
     dev: false
 
+  /rehype-sanitize@6.0.0:
+    resolution: {integrity: sha512-CsnhKNsyI8Tub6L4sm5ZFsme4puGfc6pYylvXo1AeqaGbjOYyzNv3qZPwvs0oMJ39eryyeOdmxwUIo94IpEhqg==}
+    dependencies:
+      '@types/hast': 3.0.4
+      hast-util-sanitize: 5.0.2
+    dev: false
+
   /rehype-slug@6.0.0:
     resolution: {integrity: sha512-lWyvf/jwu+oS5+hL5eClVd3hNdmwM1kAC0BUvEGD19pajQMIzcNUd/k9GsfQ+FfECvX+JE+e9/btsKH0EjJT6A==}
     dependencies:

web/src/app/home/bots/components/bot-form/BotForm.tsx

@@ -1,4 +1,4 @@
-import React, { useEffect, useState } from 'react';
+import React, { useEffect, useMemo, useState } from 'react';
 import {
   IChooseAdapterEntity,
   IPipelineEntity,
@@ -113,109 +113,73 @@ export default function BotForm({
   const [dynamicFormConfigList, setDynamicFormConfigList] = useState<
     IDynamicFormItemSchema[]
   >([]);
-  const [filteredDynamicFormConfigList, setFilteredDynamicFormConfigList] =
-    useState<IDynamicFormItemSchema[]>([]);
   const [, setIsLoading] = useState<boolean>(false);
   const [webhookUrl, setWebhookUrl] = useState<string>('');
-  const webhookInputRef = React.useRef<HTMLInputElement>(null);
+  const [extraWebhookUrl, setExtraWebhookUrl] = useState<string>('');
   const [copied, setCopied] = useState<boolean>(false);
+  const [extraCopied, setExtraCopied] = useState<boolean>(false);
 
   // Watch adapter and adapter_config for filtering
   const currentAdapter = form.watch('adapter');
   const currentAdapterConfig = form.watch('adapter_config');
 
+  // Derive the filtered config list via useMemo instead of useEffect+setState
+  // to avoid creating new array references that would cause DynamicFormComponent
+  // to re-subscribe its form.watch, re-emit values, and trigger an infinite loop.
+  // Only depend on the specific field we care about (enable-webhook) rather than
+  // the entire currentAdapterConfig object, which changes on every emission.
+  const enableWebhook = currentAdapterConfig?.['enable-webhook'];
+  const filteredDynamicFormConfigList = useMemo(() => {
+    if (currentAdapter === 'lark' && enableWebhook === false) {
+      // Hide encrypt-key field when webhook is disabled
+      return dynamicFormConfigList.filter(
+        (config) => config.name !== 'encrypt-key',
+      );
+    }
+    // For non-Lark adapters or when webhook is enabled/undefined, show all fields
+    return dynamicFormConfigList;
+  }, [currentAdapter, enableWebhook, dynamicFormConfigList]);
+
   useEffect(() => {
     setBotFormValues();
   }, []);
 
-  // Filter dynamic form config list based on enable-webhook status for Lark adapter
-  useEffect(() => {
-    if (currentAdapter === 'lark') {
-      const enableWebhook = currentAdapterConfig?.['enable-webhook'];
-      if (enableWebhook === false) {
-        // Hide encrypt-key field when webhook is disabled
-        setFilteredDynamicFormConfigList(
-          dynamicFormConfigList.filter(
-            (config) => config.name !== 'encrypt-key',
-          ),
-        );
-      } else {
-        // Show all fields when webhook is enabled or undefined
-        setFilteredDynamicFormConfigList(dynamicFormConfigList);
-      }
+  // 复制到剪贴板的辅助函数
+  const copyToClipboard = (
+    text: string,
+    setStatus: React.Dispatch<React.SetStateAction<boolean>>,
+  ) => {
+    if (navigator.clipboard && navigator.clipboard.writeText) {
+      navigator.clipboard
+        .writeText(text)
+        .then(() => {
+          setStatus(true);
+          setTimeout(() => setStatus(false), 2000);
+        })
+        .catch(() => {
+          // 降级：创建临时textarea复制
+          fallbackCopy(text, setStatus);
+        });
     } else {
-      // For non-Lark adapters, show all fields
-      setFilteredDynamicFormConfigList(dynamicFormConfigList);
+      fallbackCopy(text, setStatus);
     }
-  }, [currentAdapter, currentAdapterConfig, dynamicFormConfigList]);
+  };
 
-  // 复制到剪贴板的辅助函数 - 使用页面上的真实input元素
-  const copyToClipboard = () => {
-    console.log('[Copy] Attempting to copy from input element');
-
-    const inputElement = webhookInputRef.current;
-    if (!inputElement) {
-      console.error('[Copy] Input element not found');
-      return;
-    }
-
-    try {
-      // 确保input元素可见且未被禁用
-      inputElement.disabled = false;
-      inputElement.readOnly = false;
-
-      // 聚焦并选中所有文本
-      inputElement.focus();
-      inputElement.select();
-
-      // 尝试使用现代API
-      if (navigator.clipboard && navigator.clipboard.writeText) {
-        console.log(
-          '[Copy] Using Clipboard API with input value:',
-          inputElement.value,
-        );
-        navigator.clipboard
-          .writeText(inputElement.value)
-          .then(() => {
-            console.log('[Copy] Clipboard API success');
-            inputElement.blur(); // 取消选中
-            inputElement.readOnly = true;
-            setCopied(true);
-            setTimeout(() => setCopied(false), 2000);
-          })
-          .catch((err) => {
-            console.error(
-              '[Copy] Clipboard API failed, trying execCommand:',
-              err,
-            );
-            // 降级到execCommand
-            const successful = document.execCommand('copy');
-            console.log('[Copy] execCommand result:', successful);
-            inputElement.blur();
-            inputElement.readOnly = true;
-            if (successful) {
-              setCopied(true);
-              setTimeout(() => setCopied(false), 2000);
-            }
-          });
-      } else {
-        // 直接使用execCommand
-        console.log(
-          '[Copy] Using execCommand with input value:',
-          inputElement.value,
-        );
-        const successful = document.execCommand('copy');
-        console.log('[Copy] execCommand result:', successful);
-        inputElement.blur();
-        inputElement.readOnly = true;
-        if (successful) {
-          setCopied(true);
-          setTimeout(() => setCopied(false), 2000);
-        }
-      }
-    } catch (err) {
-      console.error('[Copy] Copy failed:', err);
-      inputElement.readOnly = true;
+  const fallbackCopy = (
+    text: string,
+    setStatus: React.Dispatch<React.SetStateAction<boolean>>,
+  ) => {
+    const textarea = document.createElement('textarea');
+    textarea.value = text;
+    textarea.style.position = 'fixed';
+    textarea.style.opacity = '0';
+    document.body.appendChild(textarea);
+    textarea.select();
+    const successful = document.execCommand('copy');
+    document.body.removeChild(textarea);
+    if (successful) {
+      setStatus(true);
+      setTimeout(() => setStatus(false), 2000);
     }
   };
 
@@ -240,6 +204,7 @@ export default function BotForm({
             } else {
               setWebhookUrl('');
             }
+            setExtraWebhookUrl(val.extra_webhook_full_url || '');
           })
           .catch((err) => {
             toast.error(
@@ -249,6 +214,7 @@ export default function BotForm({
       } else {
         form.reset();
         setWebhookUrl('');
+        setExtraWebhookUrl('');
       }
     });
   }
@@ -313,6 +279,7 @@ export default function BotForm({
               required: item.required,
               type: parseDynamicFormItemType(item.type),
               options: item.options,
+              show_if: item.show_if,
             }),
         ),
       );
@@ -320,14 +287,20 @@ export default function BotForm({
     setAdapterNameToDynamicConfigMap(adapterNameToDynamicConfigMap);
   }
 
-  async function getBotConfig(
-    botId: string,
-  ): Promise<z.infer<typeof formSchema> & { webhook_full_url?: string }> {
+  async function getBotConfig(botId: string): Promise<
+    z.infer<typeof formSchema> & {
+      webhook_full_url?: string;
+      extra_webhook_full_url?: string;
+    }
+  > {
     return new Promise((resolve, reject) => {
       httpClient
         .getBot(botId)
         .then((res) => {
           const bot = res.bot;
+          const runtimeValues = bot.adapter_runtime_values as
+            | Record<string, unknown>
+            | undefined;
           resolve({
             adapter: bot.adapter,
             description: bot.description,
@@ -335,10 +308,12 @@ export default function BotForm({
             adapter_config: bot.adapter_config,
             enable: bot.enable ?? true,
             use_pipeline_uuid: bot.use_pipeline_uuid ?? '',
-            webhook_full_url: bot.adapter_runtime_values
-              ? ((bot.adapter_runtime_values as Record<string, unknown>)
-                  .webhook_full_url as string)
-              : undefined,
+            webhook_full_url: runtimeValues?.webhook_full_url as
+              | string
+              | undefined,
+            extra_webhook_full_url: runtimeValues?.extra_webhook_full_url as
+              | string
+              | undefined,
           });
         })
         .catch((err) => {
@@ -529,13 +504,11 @@ export default function BotForm({
 
                 {/* Webhook 地址显示（统一 Webhook 模式） */}
                 {webhookUrl &&
-                  (currentAdapter !== 'lark' ||
-                    currentAdapterConfig?.['enable-webhook'] !== false) && (
+                  (currentAdapter !== 'lark' || enableWebhook !== false) && (
                     <FormItem>
                       <FormLabel>{t('bots.webhookUrl')}</FormLabel>
                       <div className="flex items-center gap-2">
                         <Input
-                          ref={webhookInputRef}
                           value={webhookUrl}
                           readOnly
                           className="flex-1 bg-gray-50 dark:bg-gray-900"
@@ -548,7 +521,7 @@ export default function BotForm({
                           type="button"
                           variant="outline"
                           size="sm"
-                          onClick={copyToClipboard}
+                          onClick={() => copyToClipboard(webhookUrl, setCopied)}
                         >
                           {copied ? (
                             <Check className="h-4 w-4 text-green-600 mr-2" />
@@ -558,8 +531,37 @@ export default function BotForm({
                           {t('common.copy')}
                         </Button>
                       </div>
+                      {extraWebhookUrl && (
+                        <div className="flex items-center gap-2 mt-2">
+                          <Input
+                            value={extraWebhookUrl}
+                            readOnly
+                            className="flex-1 bg-gray-50 dark:bg-gray-900"
+                            onClick={(e) => {
+                              (e.target as HTMLInputElement).select();
+                            }}
+                          />
+                          <Button
+                            type="button"
+                            variant="outline"
+                            size="sm"
+                            onClick={() =>
+                              copyToClipboard(extraWebhookUrl, setExtraCopied)
+                            }
+                          >
+                            {extraCopied ? (
+                              <Check className="h-4 w-4 text-green-600 mr-2" />
+                            ) : (
+                              <Copy className="h-4 w-4 mr-2" />
+                            )}
+                            {t('common.copy')}
+                          </Button>
+                        </div>
+                      )}
                       <p className="text-sm text-gray-500 mt-1">
-                        {t('bots.webhookUrlHint')}
+                        {extraWebhookUrl
+                          ? t('bots.webhookUrlHintEither')
+                          : t('bots.webhookUrlHint')}
                       </p>
                     </FormItem>
                   )}
@@ -666,7 +668,7 @@ export default function BotForm({
                 </div>
                 <DynamicFormComponent
                   itemConfigList={filteredDynamicFormConfigList}
-                  initialValues={form.watch('adapter_config')}
+                  initialValues={currentAdapterConfig}
                   onSubmit={(values) => {
                     form.setValue('adapter_config', values);
                   }}

web/src/app/home/bots/components/bot-session/BotSessionMonitor.tsx

@@ -6,6 +6,7 @@ import { httpClient } from '@/app/infra/http/HttpClient';
 import { ScrollArea } from '@/components/ui/scroll-area';
 import { Button } from '@/components/ui/button';
 import { cn } from '@/lib/utils';
+import { Copy, Check } from 'lucide-react';
 import {
   MessageChainComponent,
   Plain,
@@ -27,6 +28,7 @@ interface SessionInfo {
   is_active: boolean;
   platform?: string | null;
   user_id?: string | null;
+  user_name?: string | null;
 }
 
 interface SessionMessage {
@@ -60,8 +62,29 @@ export default function BotSessionMonitor({ botId }: BotSessionMonitorProps) {
   const [messages, setMessages] = useState<SessionMessage[]>([]);
   const [loadingSessions, setLoadingSessions] = useState(false);
   const [loadingMessages, setLoadingMessages] = useState(false);
+  const [copiedUserId, setCopiedUserId] = useState(false);
   const messagesContainerRef = useRef<HTMLDivElement>(null);
 
+  const parseSessionType = (sessionId: string): string | null => {
+    const idx = sessionId.indexOf('_');
+    if (idx === -1) return null;
+    const type = sessionId.slice(0, idx);
+    if (type === 'person' || type === 'group') return type;
+    return null;
+  };
+
+  const abbreviateId = (id: string): string => {
+    if (id.length <= 10) return id;
+    return `${id.slice(0, 4)}..${id.slice(-4)}`;
+  };
+
+  const copyUserId = (userId: string) => {
+    navigator.clipboard.writeText(userId).then(() => {
+      setCopiedUserId(true);
+      setTimeout(() => setCopiedUserId(false), 2000);
+    });
+  };
+
   const loadSessions = useCallback(async () => {
     setLoadingSessions(true);
     try {
@@ -338,24 +361,36 @@ export default function BotSessionMonitor({ botId }: BotSessionMonitorProps) {
                   >
                     <div className="flex items-center justify-between mb-0.5">
                       <span className="text-sm font-medium truncate mr-2">
-                        {session.user_id || session.session_id.slice(0, 12)}
+                        {session.user_name ||
+                          session.user_id ||
+                          session.session_id.slice(0, 12)}
                       </span>
                       <span className="text-[11px] text-muted-foreground tabular-nums flex-shrink-0">
                         {formatRelativeTime(session.last_activity)}
                       </span>
                     </div>
                     <div className="flex items-center gap-1.5 text-xs text-muted-foreground">
+                      {parseSessionType(session.session_id) && (
+                        <span className="px-1 py-0.5 rounded bg-muted text-[10px]">
+                          {parseSessionType(session.session_id)}
+                        </span>
+                      )}
                       {session.platform && (
                         <span className="px-1 py-0.5 rounded bg-muted text-[10px]">
                           {session.platform}
                         </span>
                       )}
+                      {session.user_id && (
+                        <span className="truncate text-[10px]">
+                          {abbreviateId(session.user_id)}
+                        </span>
+                      )}
                       {session.is_active && (
                         <span className="flex items-center gap-0.5 text-green-600 dark:text-green-400">
                           <span className="w-1.5 h-1.5 rounded-full bg-green-500 inline-block" />
                         </span>
                       )}
-                      <span>{session.pipeline_name}</span>
+                      <span className="truncate">{session.pipeline_name}</span>
                     </div>
                   </button>
                 );
@@ -377,15 +412,42 @@ export default function BotSessionMonitor({ botId }: BotSessionMonitorProps) {
             <div className="px-6 py-3 border-b shrink-0 flex items-center justify-between">
               <div className="min-w-0">
                 <div className="text-sm font-medium truncate">
-                  {selectedSession?.user_id || selectedSessionId.slice(0, 20)}
+                  {selectedSession?.user_name ||
+                    selectedSession?.user_id ||
+                    selectedSessionId.slice(0, 20)}
                 </div>
                 <div className="flex items-center gap-2 text-xs text-muted-foreground">
+                  {parseSessionType(selectedSessionId) && (
+                    <span>{parseSessionType(selectedSessionId)}</span>
+                  )}
                   {selectedSession?.platform && (
-                    <span>{selectedSession.platform}</span>
+                    <>
+                      {parseSessionType(selectedSessionId) && <span>·</span>}
+                      <span>{selectedSession.platform}</span>
+                    </>
+                  )}
+                  {selectedSession?.user_id && (
+                    <>
+                      <span>·</span>
+                      <span className="font-mono">
+                        {selectedSession.user_id}
+                      </span>
+                      <button
+                        onClick={() => copyUserId(selectedSession.user_id!)}
+                        className="inline-flex items-center text-muted-foreground hover:text-foreground transition-colors"
+                        title={t('common.copy')}
+                      >
+                        {copiedUserId ? (
+                          <Check className="w-3 h-3 text-green-600" />
+                        ) : (
+                          <Copy className="w-3 h-3" />
+                        )}
+                      </button>
+                    </>
                   )}
                   {selectedSession?.pipeline_name && (
                     <>
-                      {selectedSession?.platform && <span>·</span>}
+                      <span>·</span>
                       <span>{selectedSession.pipeline_name}</span>
                     </>
                   )}

web/src/app/home/components/dynamic-form/DynamicFormComponent.tsx

@@ -13,20 +13,55 @@ import {
 import DynamicFormItemComponent from '@/app/home/components/dynamic-form/DynamicFormItemComponent';
 import { useEffect, useRef } from 'react';
 import { extractI18nObject } from '@/i18n/I18nProvider';
+import { useTranslation } from 'react-i18next';
 
 export default function DynamicFormComponent({
   itemConfigList,
   onSubmit,
   initialValues,
   onFileUploaded,
+  isEditing,
+  externalDependentValues,
 }: {
   itemConfigList: IDynamicFormItemSchema[];
   onSubmit?: (val: object) => unknown;
   initialValues?: Record<string, object>;
   onFileUploaded?: (fileKey: string) => void;
+  isEditing?: boolean;
+  externalDependentValues?: Record<string, unknown>;
 }) {
   const isInitialMount = useRef(true);
   const previousInitialValues = useRef(initialValues);
+  const { t } = useTranslation();
+
+  // Normalize a form value according to its field type.
+  // This ensures legacy/malformed data (e.g. a plain string for
+  // model-fallback-selector) is coerced to the expected shape
+  // so that downstream components never crash.
+  const normalizeFieldValue = (
+    item: IDynamicFormItemSchema,
+    value: unknown,
+  ): unknown => {
+    if (item.type === 'model-fallback-selector') {
+      if (value != null && typeof value === 'object' && !Array.isArray(value)) {
+        const obj = value as Record<string, unknown>;
+        return {
+          primary: typeof obj.primary === 'string' ? obj.primary : '',
+          fallbacks: Array.isArray(obj.fallbacks)
+            ? (obj.fallbacks as unknown[]).filter(
+                (v): v is string => typeof v === 'string',
+              )
+            : [],
+        };
+      }
+      // Legacy string format or any other unexpected type
+      return {
+        primary: typeof value === 'string' ? value : '',
+        fallbacks: [],
+      };
+    }
+    return value;
+  };
 
   // 根据 itemConfigList 动态生成 zod schema
   const formSchema = z.object(
@@ -55,6 +90,9 @@ export default function DynamicFormComponent({
           case 'llm-model-selector':
             fieldSchema = z.string();
             break;
+          case 'embedding-model-selector':
+            fieldSchema = z.string();
+            break;
           case 'knowledge-base-selector':
             fieldSchema = z.string();
             break;
@@ -64,6 +102,12 @@ export default function DynamicFormComponent({
           case 'bot-selector':
             fieldSchema = z.string();
             break;
+          case 'model-fallback-selector':
+            fieldSchema = z.object({
+              primary: z.string(),
+              fallbacks: z.array(z.string()),
+            });
+            break;
           case 'prompt-editor':
             fieldSchema = z.array(
               z.object({
@@ -81,7 +125,9 @@ export default function DynamicFormComponent({
           (fieldSchema instanceof z.ZodString ||
             fieldSchema instanceof z.ZodArray)
         ) {
-          fieldSchema = fieldSchema.min(1, { message: '此字段为必填项' });
+          fieldSchema = fieldSchema.min(1, {
+            message: t('common.fieldRequired'),
+          });
         }
 
         return {
@@ -99,10 +145,10 @@ export default function DynamicFormComponent({
     resolver: zodResolver(formSchema),
     defaultValues: itemConfigList.reduce((acc, item) => {
       // 优先使用 initialValues，如果没有则使用默认值
-      const value = initialValues?.[item.name] ?? item.default;
+      const rawValue = initialValues?.[item.name] ?? item.default;
       return {
         ...acc,
-        [item.name]: value,
+        [item.name]: normalizeFieldValue(item, rawValue),
       };
     }, {} as FormValues),
   });
@@ -127,7 +173,8 @@ export default function DynamicFormComponent({
       // 合并默认值和初始值
       const mergedValues = itemConfigList.reduce(
         (acc, item) => {
-          acc[item.name] = initialValues[item.name] ?? item.default;
+          const rawValue = initialValues[item.name] ?? item.default;
+          acc[item.name] = normalizeFieldValue(item, rawValue) as object;
           return acc;
         },
         {} as Record<string, object>,
@@ -141,6 +188,9 @@ export default function DynamicFormComponent({
     }
   }, [initialValues, form, itemConfigList]);
 
+  // Get reactive form values for conditional rendering
+  const watchedValues = form.watch();
+
   // Stable ref for onSubmit to avoid re-triggering the effect when the
   // parent passes a new closure on every render.
   const onSubmitRef = useRef(onSubmit);
@@ -161,6 +211,15 @@ export default function DynamicFormComponent({
     );
     onSubmitRef.current?.(initialFinalValues);
 
+    // Update previousInitialValues to the emitted snapshot so that if the
+    // parent writes these values back as new initialValues, the deep
+    // comparison in the initialValues-sync useEffect won't detect a change
+    // and won't trigger an infinite update loop.
+    previousInitialValues.current = initialFinalValues as Record<
+      string,
+      object
+    >;
+
     const subscription = form.watch(() => {
       const formValues = form.getValues();
       const finalValues = itemConfigList.reduce(
@@ -171,6 +230,7 @@ export default function DynamicFormComponent({
         {} as Record<string, object>,
       );
       onSubmitRef.current?.(finalValues);
+      previousInitialValues.current = finalValues as Record<string, object>;
     });
     return () => subscription.unsubscribe();
   }, [form, itemConfigList]);
@@ -178,34 +238,76 @@ export default function DynamicFormComponent({
   return (
     <Form {...form}>
       <div className="space-y-4">
-        {itemConfigList.map((config) => (
-          <FormField
-            key={config.id}
-            control={form.control}
-            name={config.name as keyof FormValues}
-            render={({ field }) => (
-              <FormItem>
-                <FormLabel>
-                  {extractI18nObject(config.label)}{' '}
-                  {config.required && <span className="text-red-500">*</span>}
-                </FormLabel>
-                <FormControl>
-                  <DynamicFormItemComponent
-                    config={config}
-                    field={field}
-                    onFileUploaded={onFileUploaded}
-                  />
-                </FormControl>
-                {config.description && (
-                  <p className="text-sm text-muted-foreground">
-                    {extractI18nObject(config.description)}
-                  </p>
-                )}
-                <FormMessage />
-              </FormItem>
-            )}
-          />
-        ))}
+        {itemConfigList.map((config) => {
+          if (config.show_if) {
+            const dependValue =
+              watchedValues[
+                config.show_if.field as keyof typeof watchedValues
+              ] !== undefined
+                ? watchedValues[
+                    config.show_if.field as keyof typeof watchedValues
+                  ]
+                : externalDependentValues?.[config.show_if.field];
+
+            if (
+              config.show_if.operator === 'eq' &&
+              dependValue !== config.show_if.value
+            ) {
+              return null;
+            }
+            if (
+              config.show_if.operator === 'neq' &&
+              dependValue === config.show_if.value
+            ) {
+              return null;
+            }
+            if (
+              config.show_if.operator === 'in' &&
+              Array.isArray(config.show_if.value) &&
+              !config.show_if.value.includes(dependValue)
+            ) {
+              return null;
+            }
+          }
+
+          // All fields are disabled when editing (creation_settings are immutable)
+          const isFieldDisabled = !!isEditing;
+
+          return (
+            <FormField
+              key={config.id}
+              control={form.control}
+              name={config.name as keyof FormValues}
+              render={({ field }) => (
+                <FormItem>
+                  <FormLabel>
+                    {extractI18nObject(config.label)}{' '}
+                    {config.required && <span className="text-red-500">*</span>}
+                  </FormLabel>
+                  <FormControl>
+                    <div
+                      className={
+                        isFieldDisabled ? 'pointer-events-none opacity-60' : ''
+                      }
+                    >
+                      <DynamicFormItemComponent
+                        config={config}
+                        field={field}
+                        onFileUploaded={onFileUploaded}
+                      />
+                    </div>
+                  </FormControl>
+                  {config.description && (
+                    <p className="text-sm text-muted-foreground">
+                      {extractI18nObject(config.description)}
+                    </p>
+                  )}
+                  <FormMessage />
+                </FormItem>
+              )}
+            />
+          );
+        })}
       </div>
     </Form>
   );

web/src/app/home/components/dynamic-form/DynamicFormItemComponent.tsx

@@ -22,8 +22,7 @@ import {
   LLMModel,
   Bot,
   KnowledgeBase,
-  ExternalKnowledgeBase,
-  ApiRespPluginSystemStatus,
+  EmbeddingModel,
 } from '@/app/infra/entities/api';
 import { toast } from 'sonner';
 import { useTranslation } from 'react-i18next';
@@ -51,16 +50,12 @@ export default function DynamicFormItemComponent({
   onFileUploaded?: (fileKey: string) => void;
 }) {
   const [llmModels, setLlmModels] = useState<LLMModel[]>([]);
+  const [embeddingModels, setEmbeddingModels] = useState<EmbeddingModel[]>([]);
   const [knowledgeBases, setKnowledgeBases] = useState<KnowledgeBase[]>([]);
-  const [externalKnowledgeBases, setExternalKnowledgeBases] = useState<
-    ExternalKnowledgeBase[]
-  >([]);
   const [bots, setBots] = useState<Bot[]>([]);
   const [uploading, setUploading] = useState<boolean>(false);
   const [kbDialogOpen, setKbDialogOpen] = useState(false);
   const [tempSelectedKBIds, setTempSelectedKBIds] = useState<string[]>([]);
-  const [pluginSystemStatus, setPluginSystemStatus] =
-    useState<ApiRespPluginSystemStatus | null>(null);
   const { t } = useTranslation();
 
   const handleFileUpload = async (file: File): Promise<IFileConfig | null> => {
@@ -110,6 +105,41 @@ export default function DynamicFormItemComponent({
           }
           setLlmModels(models);
         })
+        .catch((err) => {
+          toast.error(t('models.getModelListError') + err.msg);
+        });
+    }
+  }, [config.type]);
+
+  useEffect(() => {
+    if (config.type === DynamicFormItemType.EMBEDDING_MODEL_SELECTOR) {
+      httpClient
+        .getProviderEmbeddingModels()
+        .then((resp) => {
+          setEmbeddingModels(resp.models);
+        })
+        .catch((err) => {
+          toast.error(t('embedding.getModelListError') + err.msg);
+        });
+    }
+  }, [config.type]);
+
+  useEffect(() => {
+    if (config.type === DynamicFormItemType.MODEL_FALLBACK_SELECTOR) {
+      httpClient
+        .getProviderLLMModels()
+        .then((resp) => {
+          let models = resp.models;
+          if (
+            systemInfo.disable_models_service ||
+            userInfo?.account_type !== 'space'
+          ) {
+            models = models.filter(
+              (m) => m.provider?.requester !== 'space-chat-completions',
+            );
+          }
+          setLlmModels(models);
+        })
         .catch((err) => {
           toast.error('Failed to get LLM model list: ' + err.msg);
         });
@@ -127,39 +157,11 @@ export default function DynamicFormItemComponent({
           setKnowledgeBases(resp.bases);
         })
         .catch((err) => {
-          toast.error('Failed to get knowledge base list: ' + err.msg);
-        });
-
-      // Fetch plugin system status
-      httpClient
-        .getPluginSystemStatus()
-        .then((status) => {
-          setPluginSystemStatus(status);
-        })
-        .catch((err) => {
-          console.error('Failed to get plugin system status:', err);
+          toast.error(t('knowledge.getKnowledgeBaseListError') + err.msg);
         });
     }
   }, [config.type]);
 
-  useEffect(() => {
-    if (
-      (config.type === DynamicFormItemType.KNOWLEDGE_BASE_SELECTOR ||
-        config.type === DynamicFormItemType.KNOWLEDGE_BASE_MULTI_SELECTOR) &&
-      pluginSystemStatus?.is_enable &&
-      pluginSystemStatus?.is_connected
-    ) {
-      httpClient
-        .getExternalKnowledgeBases()
-        .then((resp) => {
-          setExternalKnowledgeBases(resp.bases);
-        })
-        .catch((err) => {
-          console.error('Failed to get external knowledge base list:', err);
-        });
-    }
-  }, [config.type, pluginSystemStatus]);
-
   useEffect(() => {
     if (config.type === DynamicFormItemType.BOT_SELECTOR) {
       httpClient
@@ -168,7 +170,7 @@ export default function DynamicFormItemComponent({
           setBots(resp.bots);
         })
         .catch((err) => {
-          toast.error('Failed to get bot list: ' + err.msg);
+          toast.error(t('bots.getBotListError') + err.msg);
         });
     }
   }, [config.type]);
@@ -299,7 +301,243 @@ export default function DynamicFormItemComponent({
         </Select>
       );
 
+    case DynamicFormItemType.EMBEDDING_MODEL_SELECTOR:
+      // Group embedding models by provider
+      const groupedEmbeddingModels = embeddingModels.reduce(
+        (acc, model) => {
+          const providerName = model.provider?.name || 'Unknown';
+          if (!acc[providerName]) acc[providerName] = [];
+          acc[providerName].push(model);
+          return acc;
+        },
+        {} as Record<string, EmbeddingModel[]>,
+      );
+
+      return (
+        <Select value={field.value} onValueChange={field.onChange}>
+          <SelectTrigger className="bg-[#ffffff] dark:bg-[#2a2a2e]">
+            <SelectValue placeholder={t('knowledge.selectEmbeddingModel')} />
+          </SelectTrigger>
+          <SelectContent>
+            {Object.entries(groupedEmbeddingModels).map(
+              ([providerName, models]) => (
+                <SelectGroup key={providerName}>
+                  <SelectLabel>{providerName}</SelectLabel>
+                  {models.map((model) => (
+                    <SelectItem key={model.uuid} value={model.uuid}>
+                      {model.name}
+                    </SelectItem>
+                  ))}
+                </SelectGroup>
+              ),
+            )}
+          </SelectContent>
+        </Select>
+      );
+
+    case DynamicFormItemType.MODEL_FALLBACK_SELECTOR: {
+      // Group models by provider
+      const groupedModelsForFallback = llmModels.reduce(
+        (acc, model) => {
+          const providerName =
+            model.provider?.name || model.provider?.requester || 'Unknown';
+          if (!acc[providerName]) acc[providerName] = [];
+          acc[providerName].push(model);
+          return acc;
+        },
+        {} as Record<string, LLMModel[]>,
+      );
+
+      const rawModelValue = field.value;
+      const modelValue: { primary: string; fallbacks: string[] } =
+        rawModelValue != null &&
+        typeof rawModelValue === 'object' &&
+        !Array.isArray(rawModelValue)
+          ? {
+              primary:
+                typeof (rawModelValue as Record<string, unknown>).primary ===
+                'string'
+                  ? ((rawModelValue as Record<string, unknown>)
+                      .primary as string)
+                  : '',
+              fallbacks: Array.isArray(
+                (rawModelValue as Record<string, unknown>).fallbacks,
+              )
+                ? (
+                    (rawModelValue as Record<string, unknown>)
+                      .fallbacks as unknown[]
+                  ).filter((v): v is string => typeof v === 'string')
+                : [],
+            }
+          : {
+              primary: typeof rawModelValue === 'string' ? rawModelValue : '',
+              fallbacks: [],
+            };
+
+      const renderModelSelect = (
+        value: string,
+        onChange: (val: string) => void,
+        placeholder: string,
+      ) => (
+        <Select value={value} onValueChange={onChange}>
+          <SelectTrigger className="bg-[#ffffff] dark:bg-[#2a2a2e]">
+            <SelectValue placeholder={placeholder} />
+          </SelectTrigger>
+          <SelectContent>
+            {Object.entries(groupedModelsForFallback).map(
+              ([providerName, models]) => (
+                <SelectGroup key={providerName}>
+                  <SelectLabel>{providerName}</SelectLabel>
+                  {models.map((model) => (
+                    <SelectItem key={model.uuid} value={model.uuid}>
+                      <span className="inline-flex items-center gap-1">
+                        {model.name}
+                        {model.abilities?.includes('vision') && (
+                          <Eye className="h-3 w-3 text-muted-foreground" />
+                        )}
+                        {model.abilities?.includes('func_call') && (
+                          <Wrench className="h-3 w-3 text-muted-foreground" />
+                        )}
+                      </span>
+                    </SelectItem>
+                  ))}
+                </SelectGroup>
+              ),
+            )}
+          </SelectContent>
+        </Select>
+      );
+
+      const updateValue = (patch: Partial<typeof modelValue>) => {
+        field.onChange({ ...modelValue, ...patch });
+      };
+
+      const addFallbackModel = () => {
+        updateValue({ fallbacks: [...modelValue.fallbacks, ''] });
+      };
+
+      const updateFallbackModel = (index: number, value: string) => {
+        const updated = [...modelValue.fallbacks];
+        updated[index] = value;
+        updateValue({ fallbacks: updated });
+      };
+
+      const removeFallbackModel = (index: number) => {
+        const updated = [...modelValue.fallbacks];
+        updated.splice(index, 1);
+        updateValue({ fallbacks: updated });
+      };
+
+      const moveFallbackModel = (index: number, direction: 'up' | 'down') => {
+        const updated = [...modelValue.fallbacks];
+        const newIndex = direction === 'up' ? index - 1 : index + 1;
+        if (newIndex < 0 || newIndex >= updated.length) return;
+        [updated[index], updated[newIndex]] = [
+          updated[newIndex],
+          updated[index],
+        ];
+        updateValue({ fallbacks: updated });
+      };
+
+      return (
+        <div className="space-y-3">
+          {/* Primary model selector */}
+          <div>
+            <p className="text-xs text-muted-foreground mb-1">
+              {t('models.fallback.primary')}
+            </p>
+            {renderModelSelect(
+              modelValue.primary,
+              (val) => updateValue({ primary: val }),
+              t('models.selectModel'),
+            )}
+          </div>
+
+          {/* Fallback models */}
+          {modelValue.fallbacks.length > 0 && (
+            <div className="space-y-2">
+              <p className="text-xs text-muted-foreground">
+                {t('models.fallback.fallbackList')}
+              </p>
+              {modelValue.fallbacks.map((fbUuid: string, index: number) => (
+                <div key={index} className="flex items-center gap-2">
+                  <span className="text-xs text-muted-foreground w-4 shrink-0">
+                    {index + 1}.
+                  </span>
+                  <div className="flex-1">
+                    {renderModelSelect(
+                      fbUuid,
+                      (val) => updateFallbackModel(index, val),
+                      t('models.selectModel'),
+                    )}
+                  </div>
+                  <div className="flex gap-1 shrink-0">
+                    <Button
+                      type="button"
+                      variant="ghost"
+                      size="sm"
+                      className="h-8 w-8 p-0"
+                      onClick={() => moveFallbackModel(index, 'up')}
+                      disabled={index === 0}
+                    >
+                      ↑
+                    </Button>
+                    <Button
+                      type="button"
+                      variant="ghost"
+                      size="sm"
+                      className="h-8 w-8 p-0"
+                      onClick={() => moveFallbackModel(index, 'down')}
+                      disabled={index === modelValue.fallbacks.length - 1}
+                    >
+                      ↓
+                    </Button>
+                    <Button
+                      type="button"
+                      variant="ghost"
+                      size="sm"
+                      className="h-8 w-8 p-0 text-destructive"
+                      onClick={() => removeFallbackModel(index)}
+                    >
+                      <X className="h-4 w-4" />
+                    </Button>
+                  </div>
+                </div>
+              ))}
+            </div>
+          )}
+
+          {/* Add fallback button */}
+          <Button
+            type="button"
+            variant="outline"
+            size="sm"
+            className="w-full"
+            onClick={addFallbackModel}
+          >
+            <Plus className="h-4 w-4 mr-1" />
+            {t('models.fallback.addFallback')}
+          </Button>
+        </div>
+      );
+    }
+
     case DynamicFormItemType.KNOWLEDGE_BASE_SELECTOR:
+      // Group KBs by Knowledge Engine name
+      const kbsByEngine = knowledgeBases.reduce(
+        (acc, kb) => {
+          const engineName = kb.knowledge_engine?.name
+            ? extractI18nObject(kb.knowledge_engine.name)
+            : t('knowledge.unknownEngine');
+          if (!acc[engineName]) {
+            acc[engineName] = [];
+          }
+          acc[engineName].push(kb);
+          return acc;
+        },
+        {} as Record<string, typeof knowledgeBases>,
+      );
+
       return (
         <Select value={field.value} onValueChange={field.onChange}>
           <SelectTrigger className="bg-[#ffffff] dark:bg-[#2a2a2e]">
@@ -310,53 +548,45 @@ export default function DynamicFormItemComponent({
               <SelectItem value="__none__">{t('knowledge.empty')}</SelectItem>
             </SelectGroup>
 
-            {knowledgeBases.length > 0 && (
-              <SelectGroup>
-                <SelectLabel>{t('knowledge.builtIn')}</SelectLabel>
-                {knowledgeBases.map((base) => (
+            {Object.entries(kbsByEngine).map(([engineName, kbs]) => (
+              <SelectGroup key={engineName}>
+                <SelectLabel>{engineName}</SelectLabel>
+                {kbs.map((base) => (
                   <SelectItem key={base.uuid} value={base.uuid ?? ''}>
                     {base.name}
                   </SelectItem>
                 ))}
               </SelectGroup>
-            )}
-
-            {externalKnowledgeBases.length > 0 && (
-              <SelectGroup>
-                <SelectLabel>{t('knowledge.external')}</SelectLabel>
-                {externalKnowledgeBases.map((base) => (
-                  <SelectItem key={base.uuid} value={base.uuid ?? ''}>
-                    <div className="flex items-center gap-2">
-                      <img
-                        src={httpClient.getPluginIconURL(
-                          base.plugin_author,
-                          base.plugin_name,
-                        )}
-                        alt="plugin icon"
-                        className="w-4 h-4 rounded-[8%] flex-shrink-0"
-                      />
-                      <span>{base.name}</span>
-                    </div>
-                  </SelectItem>
-                ))}
-              </SelectGroup>
-            )}
+            ))}
           </SelectContent>
         </Select>
       );
 
     case DynamicFormItemType.KNOWLEDGE_BASE_MULTI_SELECTOR:
+      // Group KBs by Knowledge Engine name for multi-selector
+      const multiKbsByEngine = knowledgeBases.reduce(
+        (acc, kb) => {
+          const engineName = kb.knowledge_engine?.name
+            ? extractI18nObject(kb.knowledge_engine.name)
+            : t('knowledge.unknownEngine');
+          if (!acc[engineName]) {
+            acc[engineName] = [];
+          }
+          acc[engineName].push(kb);
+          return acc;
+        },
+        {} as Record<string, typeof knowledgeBases>,
+      );
+
       return (
         <>
           <div className="space-y-2">
             {field.value && field.value.length > 0 ? (
               <div className="space-y-2">
                 {field.value.map((kbId: string) => {
-                  const kb = knowledgeBases.find((base) => base.uuid === kbId);
-                  const externalKb = externalKnowledgeBases.find(
+                  const currentKb = knowledgeBases.find(
                     (base) => base.uuid === kbId,
                   );
-                  const currentKb = kb || externalKb;
                   if (!currentKb) return null;
 
                   return (
@@ -365,18 +595,17 @@ export default function DynamicFormItemComponent({
                       className="flex items-center justify-between rounded-lg border p-3 hover:bg-accent"
                     >
                       <div className="flex items-center gap-2 flex-1">
-                        {externalKb && (
-                          <img
-                            src={httpClient.getPluginIconURL(
-                              externalKb.plugin_author,
-                              externalKb.plugin_name,
-                            )}
-                            alt="plugin icon"
-                            className="w-8 h-8 rounded-[8%] flex-shrink-0"
-                          />
-                        )}
                         <div className="flex-1 min-w-0">
-                          <div className="font-medium">{currentKb.name}</div>
+                          <div className="font-medium flex items-center gap-2">
+                            {currentKb.name}
+                            {currentKb.knowledge_engine?.name && (
+                              <span className="text-xs px-2 py-0.5 rounded-full bg-purple-100 text-purple-700 dark:bg-purple-900 dark:text-purple-300">
+                                {extractI18nObject(
+                                  currentKb.knowledge_engine.name,
+                                )}
+                              </span>
+                            )}
+                          </div>
                           {currentKb.description && (
                             <div className="text-sm text-muted-foreground">
                               {currentKb.description}
@@ -430,13 +659,12 @@ export default function DynamicFormItemComponent({
                 <DialogTitle>{t('knowledge.selectKnowledgeBases')}</DialogTitle>
               </DialogHeader>
               <div className="flex-1 overflow-y-auto space-y-4 pr-2">
-                {/* Built-in Knowledge Bases */}
-                {knowledgeBases.length > 0 && (
-                  <div className="space-y-2">
+                {Object.entries(multiKbsByEngine).map(([engineName, kbs]) => (
+                  <div key={engineName} className="space-y-2">
                     <div className="text-sm font-semibold text-muted-foreground px-2">
-                      {t('knowledge.builtIn')}
+                      {engineName}
                     </div>
-                    {knowledgeBases.map((base) => {
+                    {kbs.map((base) => {
                       const isSelected = tempSelectedKBIds.includes(
                         base.uuid ?? '',
                       );
@@ -469,56 +697,7 @@ export default function DynamicFormItemComponent({
                       );
                     })}
                   </div>
-                )}
-
-                {/* External Knowledge Bases */}
-                {externalKnowledgeBases.length > 0 && (
-                  <div className="space-y-2">
-                    <div className="text-sm font-semibold text-muted-foreground px-2">
-                      {t('knowledge.external')}
-                    </div>
-                    {externalKnowledgeBases.map((base) => {
-                      const isSelected = tempSelectedKBIds.includes(
-                        base.uuid ?? '',
-                      );
-                      return (
-                        <div
-                          key={base.uuid}
-                          className="flex items-center gap-3 rounded-lg border p-3 hover:bg-accent cursor-pointer"
-                          onClick={() => {
-                            const kbId = base.uuid ?? '';
-                            setTempSelectedKBIds((prev) =>
-                              prev.includes(kbId)
-                                ? prev.filter((id) => id !== kbId)
-                                : [...prev, kbId],
-                            );
-                          }}
-                        >
-                          <Checkbox
-                            checked={isSelected}
-                            aria-label={`Select ${base.name}`}
-                          />
-                          <img
-                            src={httpClient.getPluginIconURL(
-                              base.plugin_author,
-                              base.plugin_name,
-                            )}
-                            alt="plugin icon"
-                            className="w-8 h-8 rounded-[8%] flex-shrink-0"
-                          />
-                          <div className="flex-1">
-                            <div className="font-medium">{base.name}</div>
-                            {base.description && (
-                              <div className="text-sm text-muted-foreground">
-                                {base.description}
-                              </div>
-                            )}
-                          </div>
-                        </div>
-                      );
-                    })}
-                  </div>
-                )}
+                ))}
               </div>
               <DialogFooter>
                 <Button

web/src/app/home/components/dynamic-form/DynamicFormItemConfig.ts

@@ -2,6 +2,7 @@ import {
   IDynamicFormItemSchema,
   DynamicFormItemType,
   IDynamicFormItemOption,
+  IShowIfCondition,
 } from '@/app/infra/entities/form/dynamic';
 import { I18nObject } from '@/app/infra/entities/common';
 
@@ -14,6 +15,7 @@ export class DynamicFormItemConfig implements IDynamicFormItemSchema {
   type: DynamicFormItemType;
   description?: I18nObject;
   options?: IDynamicFormItemOption[];
+  show_if?: IShowIfCondition;
 
   constructor(params: IDynamicFormItemSchema) {
     this.id = params.id;
@@ -24,6 +26,7 @@ export class DynamicFormItemConfig implements IDynamicFormItemSchema {
     this.type = params.type;
     this.description = params.description;
     this.options = params.options;
+    this.show_if = params.show_if;
   }
 }
 

web/src/app/home/components/home-sidebar/HomeSidebar.tsx

@@ -422,12 +422,12 @@ export default function HomeSidebar({
                   const language = localStorage.getItem('langbot_language');
                   if (language === 'zh-Hans' || language === 'zh-Hant') {
                     window.open(
-                      'https://docs.langbot.app/zh/insight/guide.html',
+                      'https://docs.langbot.app/zh/insight/guide',
                       '_blank',
                     );
                   } else {
                     window.open(
-                      'https://docs.langbot.app/en/insight/guide.html',
+                      'https://docs.langbot.app/en/insight/guide',
                       '_blank',
                     );
                   }

web/src/app/home/components/home-sidebar/sidbarConfigList.tsx

@@ -23,9 +23,9 @@ export const sidebarConfigList = [
     route: '/home/bots',
     description: t('bots.description'),
     helpLink: {
-      en_US: 'https://docs.langbot.app/en/usage/platforms/readme.html',
-      zh_Hans: 'https://docs.langbot.app/zh/usage/platforms/readme.html',
-      ja_JP: 'https://docs.langbot.app/ja/usage/platforms/readme.html',
+      en_US: 'https://docs.langbot.app/en/usage/platforms/readme',
+      zh_Hans: 'https://docs.langbot.app/zh/usage/platforms/readme',
+      ja_JP: 'https://docs.langbot.app/ja/usage/platforms/readme',
     },
   }),
   new SidebarChildVO({
@@ -44,9 +44,9 @@ export const sidebarConfigList = [
     route: '/home/pipelines',
     description: t('pipelines.description'),
     helpLink: {
-      en_US: 'https://docs.langbot.app/en/usage/pipelines/readme.html',
-      zh_Hans: 'https://docs.langbot.app/zh/usage/pipelines/readme.html',
-      ja_JP: 'https://docs.langbot.app/ja/usage/pipelines/readme.html',
+      en_US: 'https://docs.langbot.app/en/usage/pipelines/readme',
+      zh_Hans: 'https://docs.langbot.app/zh/usage/pipelines/readme',
+      ja_JP: 'https://docs.langbot.app/ja/usage/pipelines/readme',
     },
   }),
   new SidebarChildVO({
@@ -65,8 +65,8 @@ export const sidebarConfigList = [
     route: '/home/monitoring',
     description: t('monitoring.description'),
     helpLink: {
-      en_US: 'https://docs.langbot.app/en/features/monitoring.html',
-      zh_Hans: 'https://docs.langbot.app/zh/features/monitoring.html',
+      en_US: '',
+      zh_Hans: '',
     },
   }),
   new SidebarChildVO({
@@ -84,9 +84,9 @@ export const sidebarConfigList = [
     route: '/home/knowledge',
     description: t('knowledge.description'),
     helpLink: {
-      en_US: 'https://docs.langbot.app/en/usage/knowledge/readme.html',
-      zh_Hans: 'https://docs.langbot.app/zh/usage/knowledge/readme.html',
-      ja_JP: 'https://docs.langbot.app/ja/usage/knowledge/readme.html',
+      en_US: 'https://docs.langbot.app/en/usage/knowledge/readme',
+      zh_Hans: 'https://docs.langbot.app/zh/usage/knowledge/readme',
+      ja_JP: 'https://docs.langbot.app/ja/usage/knowledge/readme',
     },
   }),
   new SidebarChildVO({
@@ -105,9 +105,9 @@ export const sidebarConfigList = [
     route: '/home/plugins',
     description: t('plugins.description'),
     helpLink: {
-      en_US: 'https://docs.langbot.app/en/usage/plugin/plugin-intro.html',
-      zh_Hans: 'https://docs.langbot.app/zh/usage/plugin/plugin-intro.html',
-      ja_JP: 'https://docs.langbot.app/ja/usage/plugin/plugin-intro.html',
+      en_US: 'https://docs.langbot.app/en/usage/plugin/plugin-intro',
+      zh_Hans: 'https://docs.langbot.app/zh/usage/plugin/plugin-intro',
+      ja_JP: 'https://docs.langbot.app/ja/usage/plugin/plugin-intro',
     },
   }),
 ];

web/src/app/home/components/models-dialog/ModelsDialog.tsx

@@ -463,14 +463,16 @@ export default function ModelsDialog({
                     )
                   : t('models.providerCount', { count: otherProviders.length })}
               </span>
-              <Button
-                size="sm"
-                variant="outline"
-                onClick={handleCreateProvider}
-              >
-                <Plus className="h-4 w-4 mr-1" />
-                {t('models.addProvider')}
-              </Button>
+              <div className="flex gap-2">
+                <Button
+                  size="sm"
+                  variant="outline"
+                  onClick={handleCreateProvider}
+                >
+                  <Plus className="h-4 w-4 mr-1" />
+                  {t('models.addProvider')}
+                </Button>
+              </div>
             </div>
 
             {/* Provider List */}

web/src/app/home/components/new-version-dialog/NewVersionDialog.tsx

@@ -4,6 +4,7 @@ import { useTranslation } from 'react-i18next';
 import ReactMarkdown from 'react-markdown';
 import remarkGfm from 'remark-gfm';
 import rehypeRaw from 'rehype-raw';
+import rehypeSanitize from 'rehype-sanitize';
 import rehypeHighlight from 'rehype-highlight';
 import i18n from 'i18next';
 import { ExternalLink } from 'lucide-react';
@@ -35,11 +36,11 @@ export default function NewVersionDialog({
   const getUpdateDocsUrl = () => {
     const language = i18n.language;
     if (language === 'zh-Hans' || language === 'zh-Hant') {
-      return 'https://docs.langbot.app/zh/deploy/update.html';
+      return 'https://docs.langbot.app/zh/deploy/update';
     } else if (language === 'ja-JP') {
-      return 'https://docs.langbot.app/ja/deploy/update.html';
+      return 'https://docs.langbot.app/ja/deploy/update';
     } else {
-      return 'https://docs.langbot.app/en/deploy/update.html';
+      return 'https://docs.langbot.app/en/deploy/update';
     }
   };
 
@@ -62,7 +63,7 @@ export default function NewVersionDialog({
           <div className="markdown-body max-w-none text-sm">
             <ReactMarkdown
               remarkPlugins={[remarkGfm]}
-              rehypePlugins={[rehypeRaw, rehypeHighlight]}
+              rehypePlugins={[rehypeRaw, rehypeSanitize, rehypeHighlight]}
               components={{
                 ul: ({ children }) => <ul className="list-disc">{children}</ul>,
                 ol: ({ children }) => (