fix(api): avoid mutating bot update payload

fix(plugin): resolve plugin page asset origin
chore: bump version 4.9.7
2026-06-04 21:06:03 +00:00 · 2026-05-16 10:54:04 +08:00 · 2026-05-14 15:39:17 +08:00 · 2026-05-12 23:48:52 +08:00 · 2026-05-12 23:47:35 +08:00 · 2026-05-12 18:28:34 +08:00
185 changed files with 13258 additions and 2624 deletions
--- a/.gitignore
+++ b/.gitignore
@@ -47,6 +47,7 @@ plugins.bak
 coverage.xml
 .coverage
 src/langbot/web/
 testsdk/
 # Build artifacts
 /dist
--- a/README.md
+++ b/README.md
@@ -84,45 +84,48 @@ docker compose up -d
 | Platform | Status | Notes |
 |----------|--------|-------|
-| Discord | ✅ |  |
+| Discord | ✅ | Official |
-| Telegram | ✅ |  |
+| Telegram | ✅ | Official |
-| Slack | ✅ |  |
+| Slack | ✅ | Official |
-| LINE | ✅ |  |
+| LINE | ✅ | Official |
-| QQ | ✅ | Personal & Official API |
+| QQ | ✅ | Personal & Official API (Channel, DM, Group) |
 | WeCom | ✅ | Enterprise WeChat, External CS, AI Bot |
 | WeChat | ✅ | Personal & Official Account |
-| Lark | ✅ |  |
+| Lark | ✅ | Official |
-| DingTalk | ✅ |  |
+| DingTalk | ✅ | Official |
-| KOOK | ✅ |  |
+| KOOK | ✅ | Official |
 | Satori | ✅ |  |
 | Email | ✅ | Matrix, Satori |
 | Matrix | ✅ | Supports multiple bridged platforms such as Signal, WhatsApp, Messenger, iMessage, Mattermost, Google Chat, IRC, XMPP, Zulip, and more |
 ---
 ## Supported LLMs & Integrations
-| Provider | Type | Status |
+| Provider                                                                                                          | Type         | Status |
-|----------|------|--------|
+| ----------------------------------------------------------------------------------------------------------------- | ------------ | ------ |
-| [OpenAI](https://platform.openai.com/) | LLM | ✅ |
+| [OpenAI](https://platform.openai.com/)                                                                            | LLM          | ✅     |
-| [Anthropic](https://www.anthropic.com/) | LLM | ✅ |
+| [Anthropic](https://www.anthropic.com/)                                                                           | LLM          | ✅     |
-| [DeepSeek](https://www.deepseek.com/) | LLM | ✅ |
+| [DeepSeek](https://www.deepseek.com/)                                                                             | LLM          | ✅     |
-| [Google Gemini](https://aistudio.google.com/prompts/new_chat) | LLM | ✅ |
+| [Google Gemini](https://aistudio.google.com/prompts/new_chat)                                                     | LLM          | ✅     |
-| [xAI](https://x.ai/) | LLM | ✅ |
+| [xAI](https://x.ai/)                                                                                              | LLM          | ✅     |
-| [Moonshot](https://www.moonshot.cn/) | LLM | ✅ |
+| [Moonshot](https://www.moonshot.cn/)                                                                              | LLM          | ✅     |
-| [Zhipu AI](https://open.bigmodel.cn/) | LLM | ✅ |
+| [Zhipu AI](https://open.bigmodel.cn/)                                                                             | LLM          | ✅     |
-| [Ollama](https://ollama.com/) | Local LLM | ✅ |
+| [Ollama](https://ollama.com/)                                                                                     | Local LLM    | ✅     |
-| [LM Studio](https://lmstudio.ai/) | Local LLM | ✅ |
+| [LM Studio](https://lmstudio.ai/)                                                                                 | Local LLM    | ✅     |
-| [Dify](https://dify.ai) | LLMOps | ✅ |
+| [Dify](https://dify.ai)                                                                                           | LLMOps       | ✅     |
-| [MCP](https://modelcontextprotocol.io/) | Protocol | ✅ |
+| [MCP](https://modelcontextprotocol.io/)                                                                           | Protocol     | ✅     |
-| [SiliconFlow](https://siliconflow.cn/) | Gateway | ✅ |
+| [SiliconFlow](https://siliconflow.cn/)                                                                            | Gateway      | ✅     |
-| [Aliyun Bailian](https://bailian.console.aliyun.com/) | Gateway | ✅ |
+| [Aliyun Bailian](https://bailian.console.aliyun.com/)                                                             | Gateway      | ✅     |
-| [Volc Engine Ark](https://console.volcengine.com/ark/region:ark+cn-beijing/model?vendor=Bytedance&view=LIST_VIEW) | Gateway | ✅ |
+| [Volc Engine Ark](https://console.volcengine.com/ark/region:ark+cn-beijing/model?vendor=Bytedance&view=LIST_VIEW) | Gateway      | ✅     |
-| [ModelScope](https://modelscope.cn/docs/model-service/API-Inference/intro) | Gateway | ✅ |
+| [ModelScope](https://modelscope.cn/docs/model-service/API-Inference/intro)                                        | Gateway      | ✅     |
-| [GiteeAI](https://ai.gitee.com/) | Gateway | ✅ |
+| [GiteeAI](https://ai.gitee.com/)                                                                                  | Gateway      | ✅     |
-| [CompShare](https://www.compshare.cn/?ytag=GPU_YY-gh_langbot) | GPU Platform | ✅ |
+| [CompShare](https://www.compshare.cn/?ytag=GPU_YY-gh_langbot)                                                     | GPU Platform | ✅     |
-| [PPIO](https://ppinfra.com/user/register?invited_by=QJKFYD&utm_source=github_langbot) | GPU Platform | ✅ |
+| [PPIO](https://ppinfra.com/user/register?invited_by=QJKFYD&utm_source=github_langbot)                             | GPU Platform | ✅     |
-| [ShengSuanYun](https://www.shengsuanyun.com/?from=CH_KYIPP758) | GPU Platform | ✅ |
+| [ShengSuanYun](https://www.shengsuanyun.com/?from=CH_KYIPP758)                                                    | GPU Platform | ✅     |
-| [接口 AI](https://jiekou.ai/) | Gateway | ✅ |
+| [接口 AI](https://jiekou.ai/)                                                                                     | Gateway      | ✅     |
-| [302.AI](https://share.302.ai/SuTG99) | Gateway | ✅ |
+| [302.AI](https://share.302.ai/SuTG99)                                                                             | Gateway      | ✅     |
 | [Qiniu](https://www.qiniu.com/ai/agent)                                                                           | Gateway      | ✅     |
 [→ View all integrations](https://link.langbot.app/en/docs/features)
@@ -130,22 +133,23 @@ docker compose up -d
 ## Why LangBot?
-| Use Case | How LangBot Helps |
+| Use Case                    | How LangBot Helps                                                                          |
-|----------|-------------------|
+| --------------------------- | ------------------------------------------------------------------------------------------ |
-| **Customer Support** | Deploy AI agents to Slack/Discord/Telegram that answer questions using your knowledge base |
+| **Customer Support**        | Deploy AI agents to Slack/Discord/Telegram that answer questions using your knowledge base |
-| **Internal Tools** | Connect n8n/Dify workflows to WeCom/DingTalk for automated business processes |
+| **Internal Tools**          | Connect n8n/Dify workflows to WeCom/DingTalk for automated business processes              |
-| **Community Management** | Moderate QQ/Discord groups with AI-powered content filtering and interaction |
+| **Community Management**    | Moderate QQ/Discord groups with AI-powered content filtering and interaction               |
-| **Multi-Platform Presence** | One bot, all platforms. Manage from a single dashboard |
+| **Multi-Platform Presence** | One bot, all platforms. Manage from a single dashboard                                     |
 ---
 ## Live Demo
 **Try it now:** https://demo.langbot.dev/
 - Email: `demo@langbot.app`
 - Password: `langbot123456`
-*Note: Public demo environment. Do not enter sensitive information.*
+_Note: Public demo environment. Do not enter sensitive information._
 ---
--- a/README_CN.md
+++ b/README_CN.md
@@ -87,13 +87,16 @@ docker compose up -d
 | QQ | ✅ | 个人号、官方机器人（频道、私聊、群聊） |
 | 微信 | ✅ | 个人微信、微信公众号 |
 | 企业微信 | ✅ | 应用消息、对外客服、智能机器人 |
-| 飞书 | ✅ |  |
+| 飞书 | ✅ | 官方 |
-| 钉钉 | ✅ |  |
+| 钉钉 | ✅ | 官方 |
-| Discord | ✅ |  |
+| Satori | ✅ |  |
-| Telegram | ✅ |  |
+| Discord | ✅ | 官方 |
-| Slack | ✅ |  |
+| Telegram | ✅ | 官方 |
-| LINE | ✅ |  |
+| Slack | ✅ | 官方 |
-| KOOK | ✅ |  |
+| LINE | ✅ | 官方 |
 | KOOK | ✅ | 官方 |
 | Email | ✅ | 只 Matrix、Satori |
 | Matrix | ✅ | 支持多种桥接平台，如 Signal、WhatsApp、Messenger、iMessage、Mattermost、Google Chat、IRC、XMPP、Zulip 等 |
 ---
@@ -124,6 +127,7 @@ docker compose up -d
 | [302.AI](https://share.302.ai/SuTG99) | 聚合平台 | ✅ |
 | [小马算力](https://www.tokenpony.cn/453z1) | 聚合平台 | ✅ |
 | [百宝箱Tbox](https://www.tbox.cn/open) | 智能体平台 | ✅ |
 | [七牛云Qiniu](https://www.qiniu.com/ai/agent) | 聚合平台 | ✅ |
 [→ 查看完整集成列表](https://link.langbot.app/zh/docs/features)
--- a/README_ES.md
+++ b/README_ES.md
@@ -83,17 +83,19 @@ docker compose up -d
 | Plataforma | Estado | Notas |
 |----------|--------|-------|
-| Discord | ✅ |  |
+| Discord | ✅ | Oficial |
-| Telegram | ✅ |  |
+| Telegram | ✅ | Oficial |
-| Slack | ✅ |  |
+| Slack | ✅ | Oficial |
-| LINE | ✅ |  |
+| LINE | ✅ | Oficial |
-| QQ | ✅ | Personal y API Oficial |
+| QQ | ✅ | Personal y API Oficial (Canal, DM, Grupo) |
 | WeCom | ✅ | WeChat Empresarial, CS Externo, AI Bot |
 | WeChat | ✅ | Personal y Cuenta Oficial |
-| Lark | ✅ |  |
+| Lark | ✅ | Oficial |
-| DingTalk | ✅ |  |
+| DingTalk | ✅ | Oficial |
-| KOOK | ✅ |  |
+| KOOK | ✅ | Oficial |
 | Satori | ✅ |  |
 | Email | ✅ | Matrix, Satori |
 | Matrix | ✅ | Admite varias plataformas puenteadas como Signal, WhatsApp, Messenger, iMessage, Mattermost, Google Chat, IRC, XMPP, Zulip y más |
 ---
@@ -122,6 +124,7 @@ docker compose up -d
 | [ShengSuanYun](https://www.shengsuanyun.com/?from=CH_KYIPP758) | Plataforma GPU | ✅ |
 | [接口 AI](https://jiekou.ai/) | Pasarela | ✅ |
 | [302.AI](https://share.302.ai/SuTG99) | Pasarela | ✅ |
 | [Qiniu](https://www.qiniu.com/ai/agent) | Pasarela | ✅ |
 [→ Ver todas las integraciones](https://link.langbot.app/en/docs/features)
--- a/README_FR.md
+++ b/README_FR.md
@@ -83,17 +83,19 @@ docker compose up -d
 | Plateforme | Statut | Notes |
 |----------|--------|-------|
-| Discord | ✅ |  |
+| Discord | ✅ | Officiel |
-| Telegram | ✅ |  |
+| Telegram | ✅ | Officiel |
-| Slack | ✅ |  |
+| Slack | ✅ | Officiel |
-| LINE | ✅ |  |
+| LINE | ✅ | Officiel |
-| QQ | ✅ | Personnel & API Officielle |
+| QQ | ✅ | Personnel & API Officielle (Canal, DM, Groupe) |
 | WeCom | ✅ | WeChat Entreprise, CS Externe, AI Bot |
 | WeChat | ✅ | Personnel & Compte Officiel |
-| Lark | ✅ |  |
+| Lark | ✅ | Officiel |
-| DingTalk | ✅ |  |
+| DingTalk | ✅ | Officiel |
-| KOOK | ✅ |  |
+| KOOK | ✅ | Officiel |
 | Satori | ✅ |  |
 | Email | ✅ | Matrix, Satori |
 | Matrix | ✅ | Prend en charge plusieurs plateformes via ponts, comme Signal, WhatsApp, Messenger, iMessage, Mattermost, Google Chat, IRC, XMPP, Zulip, etc. |
 ---
@@ -122,6 +124,7 @@ docker compose up -d
 | [CompShare](https://www.compshare.cn/?ytag=GPU_YY-gh_langbot) | Plateforme GPU | ✅ |
 | [PPIO](https://ppinfra.com/user/register?invited_by=QJKFYD&utm_source=github_langbot) | Plateforme GPU | ✅ |
 | [ShengSuanYun](https://www.shengsuanyun.com/?from=CH_KYIPP758) | Plateforme GPU | ✅ |
 | [Qiniu](https://www.qiniu.com/ai/agent) | Passerelle | ✅ |
 [→ Voir toutes les intégrations](https://link.langbot.app/en/docs/features)
--- a/README_JP.md
+++ b/README_JP.md
@@ -83,17 +83,19 @@ docker compose up -d
 | プラットフォーム | ステータス | 備考 |
 |----------|--------|-------|
-| Discord | ✅ |  |
+| Discord | ✅ | 公式 |
-| Telegram | ✅ |  |
+| Telegram | ✅ | 公式 |
-| Slack | ✅ |  |
+| Slack | ✅ | 公式 |
-| LINE | ✅ |  |
+| LINE | ✅ | 公式 |
-| QQ | ✅ | 個人 & 公式API |
+| QQ | ✅ | 個人・公式API（チャンネル・DM・グループ） |
 | WeCom | ✅ | 企業WeChat、外部CS、AIボット |
-| WeChat | ✅ | 個人 & 公式アカウント |
+| WeChat | ✅ | 個人・公式アカウント |
-| Lark | ✅ |  |
+| Lark | ✅ | 公式 |
-| DingTalk | ✅ |  |
+| DingTalk | ✅ | 公式 |
-| KOOK | ✅ |  |
+| KOOK | ✅ | 公式 |
 | Satori | ✅ |  |
 | Email | ✅ | Matrix、Satori |
 | Matrix | ✅ | Signal、WhatsApp、Messenger、iMessage、Mattermost、Google Chat、IRC、XMPP、Zulip など複数のブリッジ先プラットフォームに対応 |
 ---
@@ -122,6 +124,7 @@ docker compose up -d
 | [ShengSuanYun](https://www.shengsuanyun.com/?from=CH_KYIPP758) | GPUプラットフォーム | ✅ |
 | [接口 AI](https://jiekou.ai/) | ゲートウェイ | ✅ |
 | [302.AI](https://share.302.ai/SuTG99) | ゲートウェイ | ✅ |
 | [Qiniu](https://www.qiniu.com/ai/agent) | ゲートウェイ | ✅ |
 [→ すべての統合を表示](https://link.langbot.app/en/docs/features)
--- a/README_KO.md
+++ b/README_KO.md
@@ -83,17 +83,19 @@ docker compose up -d
 | 플랫폼 | 상태 | 비고 |
 |--------|------|------|
-| Discord | ✅ |  |
+| Discord | ✅ | 공식 |
-| Telegram | ✅ |  |
+| Telegram | ✅ | 공식 |
-| Slack | ✅ |  |
+| Slack | ✅ | 공식 |
-| LINE | ✅ |  |
+| LINE | ✅ | 공식 |
-| QQ | ✅ | 개인 및 공식 API |
+| QQ | ✅ | 개인 및 공식 API (채널, DM, 그룹) |
 | WeCom | ✅ | 기업 WeChat, 외부 CS, AI Bot |
 | WeChat | ✅ | 개인 및 공식 계정 |
-| Lark | ✅ |  |
+| Lark | ✅ | 공식 |
-| DingTalk | ✅ |  |
+| DingTalk | ✅ | 공식 |
-| KOOK | ✅ |  |
+| KOOK | ✅ | 공식 |
 | Satori | ✅ |  |
 | Email | ✅ | Matrix, Satori |
 | Matrix | ✅ | Signal, WhatsApp, Messenger, iMessage, Mattermost, Google Chat, IRC, XMPP, Zulip 등 여러 브리지 플랫폼 지원 |
 ---
@@ -122,6 +124,7 @@ docker compose up -d
 | [ShengSuanYun](https://www.shengsuanyun.com/?from=CH_KYIPP758) | GPU 플랫폼 | ✅ |
 | [接口 AI](https://jiekou.ai/) | 게이트웨이 | ✅ |
 | [302.AI](https://share.302.ai/SuTG99) | 게이트웨이 | ✅ |
 | [Qiniu](https://www.qiniu.com/ai/agent) | 게이트웨이 | ✅ |
 [→ 모든 통합 보기](https://link.langbot.app/en/docs/features)
--- a/README_RU.md
+++ b/README_RU.md
@@ -83,17 +83,19 @@ docker compose up -d
 | Платформа | Статус | Примечания |
 |-----------|--------|------------|
-| Discord | ✅ |  |
+| Discord | ✅ | Официальный |
-| Telegram | ✅ |  |
+| Telegram | ✅ | Официальный |
-| Slack | ✅ |  |
+| Slack | ✅ | Официальный |
-| LINE | ✅ |  |
+| LINE | ✅ | Официальный |
-| QQ | ✅ | Личный и официальный API |
+| QQ | ✅ | Личный и официальный API (Канал, ЛС, Группа) |
 | WeCom | ✅ | Корпоративный WeChat, внешний CS, AI-бот |
 | WeChat | ✅ | Личный и официальный аккаунт |
-| Lark | ✅ |  |
+| Lark | ✅ | Официальный |
-| DingTalk | ✅ |  |
+| DingTalk | ✅ | Официальный |
-| KOOK | ✅ |  |
+| KOOK | ✅ | Официальный |
 | Satori | ✅ |  |
 | Email | ✅ | Matrix, Satori |
 | Matrix | ✅ | Поддерживает несколько платформ через мосты, включая Signal, WhatsApp, Messenger, iMessage, Mattermost, Google Chat, IRC, XMPP, Zulip и другие |
 ---
@@ -122,6 +124,7 @@ docker compose up -d
 | [CompShare](https://www.compshare.cn/?ytag=GPU_YY-gh_langbot) | Платформа GPU | ✅ |
 | [PPIO](https://ppinfra.com/user/register?invited_by=QJKFYD&utm_source=github_langbot) | Платформа GPU | ✅ |
 | [ShengSuanYun](https://www.shengsuanyun.com/?from=CH_KYIPP758) | Платформа GPU | ✅ |
 | [Qiniu](https://www.qiniu.com/ai/agent) | Шлюз | ✅ |
 [→ Смотреть все интеграции](https://link.langbot.app/en/docs/features)
--- a/README_TW.md
+++ b/README_TW.md
@@ -85,17 +85,19 @@ docker compose up -d
 | 平台 | 狀態 | 備註 |
 |------|------|------|
 | Discord | ✅ | 官方 |
 | Telegram | ✅ | 官方 |
 | Slack | ✅ | 官方 |
 | LINE | ✅ | 官方 |
 | QQ | ✅ | 個人號、官方機器人（頻道、私聊、群聊） |
 | 微信 | ✅ | 個人微信、微信公眾號 |
 | 企業微信 | ✅ | 應用訊息、對外客服、智能機器人 |
-| 飛書 | ✅ |  |
+| 微信 | ✅ | 個人微信、微信公眾號 |
-| 釘釘 | ✅ |  |
+| 飛書 | ✅ | 官方 |
-| Discord | ✅ |  |
+| 釘釘 | ✅ | 官方 |
-| Telegram | ✅ |  |
+| KOOK | ✅ | 官方 |
 | Slack | ✅ |  |
 | LINE | ✅ |  |
 | KOOK | ✅ |  |
 | Satori | ✅ |  |
 | Email | ✅ | 只 Matrix、Satori |
 | Matrix | ✅ | 支援多種橋接平台，如 Signal、WhatsApp、Messenger、iMessage、Mattermost、Google Chat、IRC、XMPP、Zulip 等 |
 ---
@@ -124,6 +126,7 @@ docker compose up -d
 | [PPIO](https://ppinfra.com/user/register?invited_by=QJKFYD&utm_source=github_langbot) | GPU 平台 | ✅ |
 | [接口 AI](https://jiekou.ai/) | 聚合平台 | ✅ |
 | [302.AI](https://share.302.ai/SuTG99) | 聚合平台 | ✅ |
 | [Qiniu](https://www.qiniu.com/ai/agent) | 聚合平台 | ✅ |
 ### TTS（語音合成）
--- a/README_VI.md
+++ b/README_VI.md
@@ -83,17 +83,19 @@ docker compose up -d
 | Nền tảng | Trạng thái | Ghi chú |
 |----------|--------|-------|
-| Discord | ✅ |  |
+| Discord | ✅ | Chính thức |
-| Telegram | ✅ |  |
+| Telegram | ✅ | Chính thức |
-| Slack | ✅ |  |
+| Slack | ✅ | Chính thức |
-| LINE | ✅ |  |
+| LINE | ✅ | Chính thức |
-| QQ | ✅ | Cá nhân & API chính thức |
+| QQ | ✅ | Cá nhân & API chính thức (Kênh, DM, Nhóm) |
 | WeCom | ✅ | WeChat doanh nghiệp, CS bên ngoài, AI Bot |
 | WeChat | ✅ | Cá nhân & Tài khoản công khai |
-| Lark | ✅ |  |
+| Lark | ✅ | Chính thức |
-| DingTalk | ✅ |  |
+| DingTalk | ✅ | Chính thức |
-| KOOK | ✅ |  |
+| KOOK | ✅ | Chính thức |
 | Satori | ✅ |  |
 | Email | ✅ | Matrix, Satori |
 | Matrix | ✅ | Hỗ trợ nhiều nền tảng qua bridge như Signal, WhatsApp, Messenger, iMessage, Mattermost, Google Chat, IRC, XMPP, Zulip và hơn thế nữa |
 ---
@@ -122,6 +124,7 @@ docker compose up -d
 | [ShengSuanYun](https://www.shengsuanyun.com/?from=CH_KYIPP758) | Nền tảng GPU | ✅ |
 | [接口 AI](https://jiekou.ai/) | Cổng | ✅ |
 | [302.AI](https://share.302.ai/SuTG99) | Cổng | ✅ |
 | [Qiniu](https://www.qiniu.com/ai/agent) | Cổng | ✅ |
 [→ Xem tất cả tích hợp](https://link.langbot.app/en/docs/features)
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -1,6 +1,6 @@
 [project]
 name = "langbot"
-version = "4.9.6"
+version = "4.9.7"
 description = "Production-grade platform for building agentic IM bots"
 readme = "README.md"
 license-files = ["LICENSE"]
@@ -22,7 +22,7 @@ dependencies = [
    "discord-py>=2.5.2",
    "pynacl>=1.5.0", # Required for Discord voice support
    "gewechat-client>=0.1.5",
-    "lark-oapi>=1.4.15",
+    "lark-oapi>=1.5.5",
    "mcp>=1.25.0",
    "nakuru-project-idk>=0.0.2.1",
    "ollama>=0.4.8",
@@ -35,6 +35,7 @@ dependencies = [
    "python-telegram-bot>=22.0",
    "pyyaml>=6.0.2",
    "qq-botpy-rc>=1.2.1.6",
    "qrcode>=7.4",
    "quart>=0.20.0",
    "quart-cors>=0.8.0",
    "requests>=2.32.3",
@@ -69,15 +70,15 @@ dependencies = [
    "chromadb>=1.0.0,<2.0.0",
    "qdrant-client (>=1.15.1,<2.0.0)",
    "pyseekdb==1.1.0.post3",
-    "langbot-plugin==0.3.8",
+    "langbot-plugin==0.3.11",
    "asyncpg>=0.30.0",
    "line-bot-sdk>=3.19.0",
    "matrix-nio>=0.25.2",
    "tboxsdk>=0.0.10",
    "boto3>=1.35.0",
    "pymilvus>=2.6.4",
    "pgvector>=0.4.1",
    "botocore>=1.42.39",
    "litellm>=1.0.0",
 ]
 keywords = [
    "bot",
--- a/src/langbot/init.py
+++ b/src/langbot/init.py
@@ -1,3 +1,3 @@
 """LangBot - Production-grade platform for building agentic IM bots"""
-__version__ = '4.9.6'
+__version__ = '4.9.7'
--- a/src/langbot/libs/dingtalk_api/api.py
+++ b/src/langbot/libs/dingtalk_api/api.py
@@ -481,6 +481,12 @@ class DingTalkClient:
        card_data['config'] = json.dumps({'autoLayout': card_auto_layout})
        card_data['content'] = ''
        # 将用户的消息内容作为卡片的查询参数，方便后续处理
        if incoming_message.message_type == 'text':
            card_data['query'] = incoming_message.get_text_list()[0]
        else:
            card_data['query'] = '...'
        card_instance = dingtalk_stream.AICardReplier(self.client, incoming_message)
        # print(card_instance)
        # 先投放卡片: https://open.dingtalk.com/document/orgapp/create-and-deliver-cards
--- a/src/langbot/libs/qq_official_api/api.py
+++ b/src/langbot/libs/qq_official_api/api.py
@@ -1,8 +1,10 @@
 import re
 import time
 import asyncio
 from quart import request
 import httpx
 from quart import Quart
-from typing import Callable, Dict, Any
+from typing import Callable, Dict, Any, Optional
 import langbot_plugin.api.entities.builtin.platform.events as platform_events
 from .qqofficialevent import QQOfficialEvent
 import json
@@ -32,6 +34,8 @@ class QQOfficialClient:
        self.access_token = ''
        self.access_token_expiry_time = None
        self.logger = logger
        self._msg_seq_counter = 0
        self._token_refresh_task: Optional[asyncio.Task] = None
    async def check_access_token(self):
        """检查access_token是否存在"""
@@ -50,18 +54,18 @@ class QQOfficialClient:
            headers = {
                'content-type': 'application/json',
            }
-            try:
+            response = await client.post(url, json=params, headers=headers)
-                response = await client.post(url, json=params, headers=headers)
+            if response.status_code != 200:
-                if response.status_code == 200:
+                raise Exception(f'Failed to get access_token: HTTP {response.status_code} {response.text}')
-                    response_data = response.json()
+            response_data = response.json()
-                access_token = response_data.get('access_token')
+            access_token = response_data.get('access_token')
-                expires_in = int(response_data.get('expires_in', 7200))
+            expires_in = int(response_data.get('expires_in', 7200))
-                self.access_token_expiry_time = time.time() + expires_in - 60
+            self.access_token_expiry_time = time.time() + expires_in - 60
-                if access_token:
+            if access_token:
-                    self.access_token = access_token
+                self.access_token = access_token
-            except Exception as e:
+                await self.logger.info(f'access_token obtained, expires_in={expires_in}s')
-                await self.logger.error(f'获取access_token失败: {response_data}')
+            else:
-                raise Exception(f'获取access_token失败: {e}')
+                raise Exception('Failed to get access_token: no access_token in response')
    async def handle_callback_request(self):
        """处理回调请求（独立端口模式，使用全局 request）"""
@@ -87,10 +91,10 @@ class QQOfficialClient:
        try:
            body = await req.get_data()
-            print(f'[QQ Official] Received request, body length: {len(body)}')
+            await self.logger.info(f'Received request, body length: {len(body)}')
            if not body or len(body) == 0:
-                print('[QQ Official] Received empty body, might be health check or GET request')
+                await self.logger.info('Received empty body, might be health check or GET request')
                return {'code': 0, 'message': 'ok'}, 200
            payload = json.loads(body)
@@ -111,7 +115,6 @@ class QQOfficialClient:
            return {'code': 0, 'message': 'success'}
        except Exception as e:
            print(f'[QQ Official] ERROR: {traceback.format_exc()}')
            await self.logger.error(f'Error in handle_callback_request: {traceback.format_exc()}')
            return {'error': str(e)}, 400
@@ -139,21 +142,24 @@ class QQOfficialClient:
    async def get_message(self, msg: dict) -> Dict[str, Any]:
        """获取消息"""
        d = msg.get('d', {})
        if not isinstance(d, dict):
            return {}
        message_data = {
            't': msg.get('t', {}),
-            'user_openid': msg.get('d', {}).get('author', {}).get('user_openid', {}),
+            'user_openid': d.get('author', {}).get('user_openid', {}),
-            'timestamp': msg.get('d', {}).get('timestamp', {}),
+            'timestamp': d.get('timestamp', {}),
-            'd_author_id': msg.get('d', {}).get('author', {}).get('id', {}),
+            'd_author_id': d.get('author', {}).get('id', {}),
-            'content': msg.get('d', {}).get('content', {}),
+            'content': d.get('content', {}),
-            'd_id': msg.get('d', {}).get('id', {}),
+            'd_id': d.get('id', {}),
            'id': msg.get('id', {}),
-            'channel_id': msg.get('d', {}).get('channel_id', {}),
+            'channel_id': d.get('channel_id', {}),
-            'username': msg.get('d', {}).get('author', {}).get('username', {}),
+            'username': d.get('author', {}).get('username', {}),
-            'guild_id': msg.get('d', {}).get('guild_id', {}),
+            'guild_id': d.get('guild_id', {}),
-            'member_openid': msg.get('d', {}).get('author', {}).get('openid', {}),
+            'member_openid': d.get('author', {}).get('openid', {}),
-            'group_openid': msg.get('d', {}).get('group_openid', {}),
+            'group_openid': d.get('group_openid', {}),
        }
-        attachments = msg.get('d', {}).get('attachments', [])
+        attachments = d.get('attachments', [])
        image_attachments = [attachment['url'] for attachment in attachments if await self.is_image(attachment)]
        image_attachments_type = [
            attachment['content_type'] for attachment in attachments if await self.is_image(attachment)
@@ -192,7 +198,7 @@ class QQOfficialClient:
            if response.status_code == 200:
                return
            else:
-                await self.logger.error(f'发送私聊消息失败: {response_data}')
+                await self.logger.error(f'Failed to send private message: {response_data}')
                raise ValueError(response)
    async def send_group_text_msg(self, group_openid: str, content: str, msg_id: str):
@@ -215,7 +221,7 @@ class QQOfficialClient:
            if response.status_code == 200:
                return
            else:
-                await self.logger.error(f'发送群聊消息失败:{response.json()}')
+                await self.logger.error(f'Failed to send group message: {response.json()}')
                raise Exception(response.read().decode())
    async def send_channle_group_text_msg(self, channel_id: str, content: str, msg_id: str):
@@ -238,7 +244,7 @@ class QQOfficialClient:
            if response.status_code == 200:
                return True
            else:
-                await self.logger.error(f'发送频道群聊消息失败: {response.json()}')
+                await self.logger.error(f'Failed to send channel group message: {response.json()}')
                raise Exception(response)
    async def send_channle_private_text_msg(self, guild_id: str, content: str, msg_id: str):
@@ -261,9 +267,224 @@ class QQOfficialClient:
            if response.status_code == 200:
                return True
            else:
-                await self.logger.error(f'发送频道私聊消息失败: {response.json()}')
+                await self.logger.error(f'Failed to send channel private message: {response.json()}')
                raise Exception(response)
    # ---- 富媒体消息 ----
    # 媒体文件类型
    MEDIA_TYPE_IMAGE = 1
    MEDIA_TYPE_VIDEO = 2
    MEDIA_TYPE_VOICE = 3
    MEDIA_TYPE_FILE = 4
    async def upload_media(
        self,
        target_type: str,
        target_id: str,
        file_type: int,
        file_url: str = None,
        file_data: str = None,
        file_name: str = None,
    ) -> str:
        """上传媒体文件，返回 file_info。
        Args:
            target_type: 'c2c' | 'group'
            target_id: 用户 openid 或群 openid
            file_type: 1=图片, 2=视频, 3=语音, 4=文件
            file_url: 在线 URL（与 file_data 二选一）
            file_data: base64 编码的文件数据或 data URL（与 file_url 二选一）
            file_name: 文件名（file_type=4 时必填）
        """
        if not await self.check_access_token():
            await self.get_access_token()
        if target_type == 'c2c':
            url = f'{self.base_url}/v2/users/{target_id}/files'
        elif target_type == 'group':
            url = f'{self.base_url}/v2/groups/{target_id}/files'
        else:
            raise ValueError(f'Unsupported target_type: {target_type}')
        body = {
            'file_type': file_type,
            'srv_send_msg': False,
        }
        if file_url:
            body['url'] = file_url
        elif file_data:
            # 处理 data URL 格式: data:image/png;base64,xxxxx
            if file_data.startswith('data:'):
                match = re.match(r'^data:[^;]+;base64,(.+)$', file_data, re.DOTALL)
                if match:
                    body['file_data'] = match.group(1)
                else:
                    body['file_data'] = file_data
            else:
                body['file_data'] = file_data
        else:
            raise ValueError('file_url or file_data is required')
        if file_type == self.MEDIA_TYPE_FILE and file_name:
            body['file_name'] = file_name
        async with httpx.AsyncClient(timeout=120) as client:
            headers = {
                'Authorization': f'QQBot {self.access_token}',
                'Content-Type': 'application/json',
            }
            response = await client.post(url, headers=headers, json=body)
            if response.status_code == 200:
                data = response.json()
                file_info = data.get('file_info', '')
                preview = file_info[:80] + '...' if len(file_info) > 80 else file_info
                await self.logger.info(f'Upload media success, file_info={preview}')
                return file_info
            else:
                raise Exception(f'Failed to upload media: HTTP {response.status_code} {response.text}')
    async def _send_media_msg(
        self,
        target_type: str,
        target_id: str,
        file_info: str,
        msg_id: str = None,
        content: str = None,
    ):
        """发送富媒体消息（msg_type=7）"""
        if not await self.check_access_token():
            await self.get_access_token()
        if target_type == 'c2c':
            url = f'{self.base_url}/v2/users/{target_id}/messages'
        elif target_type == 'group':
            url = f'{self.base_url}/v2/groups/{target_id}/messages'
        else:
            raise ValueError(f'Unsupported target_type: {target_type}')
        self._msg_seq_counter += 1
        msg_seq = self._msg_seq_counter
        body = {
            'msg_type': 7,
            'media': {'file_info': file_info},
            'msg_seq': msg_seq,
        }
        if content:
            body['content'] = content
        if msg_id:
            body['msg_id'] = msg_id
        async with httpx.AsyncClient(timeout=120) as client:
            headers = {
                'Authorization': f'QQBot {self.access_token}',
                'Content-Type': 'application/json',
            }
            await self.logger.info(f'Sending rich media: {json.dumps(body, ensure_ascii=False)[:200]}')
            response = await client.post(url, headers=headers, json=body)
            if response.status_code != 200:
                raise Exception(f'Failed to send rich media message: HTTP {response.status_code} {response.text}')
    async def send_image_msg(
        self,
        target_type: str,
        target_id: str,
        file_url: str = None,
        file_data: str = None,
        msg_id: str = None,
        content: str = None,
    ):
        """发送图片消息"""
        file_info = await self.upload_media(
            target_type,
            target_id,
            self.MEDIA_TYPE_IMAGE,
            file_url=file_url,
            file_data=file_data,
        )
        await self._send_media_msg(target_type, target_id, file_info, msg_id, content)
    async def send_voice_msg(
        self,
        target_type: str,
        target_id: str,
        file_url: str = None,
        file_data: str = None,
        msg_id: str = None,
    ):
        """发送语音消息"""
        file_info = await self.upload_media(
            target_type,
            target_id,
            self.MEDIA_TYPE_VOICE,
            file_url=file_url,
            file_data=file_data,
        )
        await self._send_media_msg(target_type, target_id, file_info, msg_id)
    async def send_file_msg(
        self,
        target_type: str,
        target_id: str,
        file_url: str = None,
        file_data: str = None,
        file_name: str = None,
        msg_id: str = None,
    ):
        """发送文件消息（含视频）"""
        file_info = await self.upload_media(
            target_type,
            target_id,
            self.MEDIA_TYPE_FILE,
            file_url=file_url,
            file_data=file_data,
            file_name=file_name,
        )
        await self._send_media_msg(target_type, target_id, file_info, msg_id)
    async def send_stream_msg(
        self,
        user_openid: str,
        content: str,
        event_id: str,
        msg_id: str,
        msg_seq: int = 1,
        index: int = 0,
        stream_msg_id: str = None,
        input_state: int = 1,
    ):
        """发送流式消息（C2C 私聊）。
        Args:
            input_state: 1=生成中, 10=生成结束
        """
        if not await self.check_access_token():
            await self.get_access_token()
        url = f'{self.base_url}/v2/users/{user_openid}/stream_messages'
        body = {
            'input_mode': 'replace',
            'input_state': input_state,
            'content_type': 'markdown',
            'content_raw': content,
            'event_id': event_id,
            'msg_id': msg_id,
            'msg_seq': msg_seq,
            'index': index,
        }
        if stream_msg_id:
            body['stream_msg_id'] = stream_msg_id
        async with httpx.AsyncClient(timeout=120) as client:
            headers = {
                'Authorization': f'QQBot {self.access_token}',
                'Content-Type': 'application/json',
            }
            response = await client.post(url, headers=headers, json=body)
            if response.status_code != 200:
                raise Exception(f'Failed to send stream message: HTTP {response.status_code} {response.text}')
            return response.json()
    async def is_token_expired(self):
        """检查token是否过期"""
        if self.access_token_expiry_time is None:
@@ -292,3 +513,325 @@ class QQOfficialClient:
            'signature': signature,
        }
        return response
    # ---- WebSocket Gateway ----
    # Reference: https://bot.q.qq.com/wiki/develop/api-v2/dev-prepare/interface-framework/event-emit.html
    INTENT_GUILDS = 1 << 0
    INTENT_GUILD_MEMBERS = 1 << 1
    INTENT_PUBLIC_GUILD_MESSAGES = 1 << 30
    INTENT_DIRECT_MESSAGE = 1 << 12
    INTENT_GROUP_AND_C2C = 1 << 25
    INTENT_INTERACTION = 1 << 26
    FULL_INTENTS = (
        INTENT_GUILDS
        | INTENT_GUILD_MEMBERS
        | INTENT_PUBLIC_GUILD_MESSAGES
        | INTENT_DIRECT_MESSAGE
        | INTENT_GROUP_AND_C2C
        | INTENT_INTERACTION
    )
    async def get_gateway_url(self) -> str:
        """获取 WebSocket 网关地址"""
        if not await self.check_access_token():
            await self.get_access_token()
        url = f'{self.base_url}/gateway'
        async with httpx.AsyncClient() as client:
            headers = {
                'Authorization': f'QQBot {self.access_token}',
            }
            response = await client.get(url, headers=headers)
            if response.status_code == 200:
                data = response.json()
                ws_url = data.get('url', '')
                if not ws_url:
                    raise Exception('Gateway URL is empty')
                return ws_url
            else:
                raise Exception(f'Failed to get Gateway URL: HTTP {response.status_code} {response.text}')
    async def _background_token_refresh(self):
        """在 token 到期前主动刷新"""
        try:
            while True:
                if self.access_token_expiry_time:
                    remain = self.access_token_expiry_time - time.time()
                    if remain > 120:
                        await asyncio.sleep(remain - 60)
                        continue
                self.access_token = ''
                self.access_token_expiry_time = None
                if await self.check_access_token():
                    await asyncio.sleep(60)
                else:
                    await self.get_access_token()
                    await asyncio.sleep(60)
        except asyncio.CancelledError:
            pass
    async def connect_gateway(
        self,
        on_event: Callable[[str, dict], Any],
        on_ready: Optional[Callable[[], Any]] = None,
        on_error: Optional[Callable[[Exception], Any]] = None,
    ):
        """WebSocket 网关连接，含重连逻辑。持续重连直到达到最大次数或被取消。
        Args:
            on_event: 收到 op=0 Dispatch 事件时的回调，参数为 (event_type, event_data)
            on_ready: 连接就绪 (收到 READY) 时的回调
            on_error: 发生错误时的回调
        """
        import websockets
        session_id = ''
        last_seq = 0
        reconnect_attempts = 0
        max_reconnect_attempts = 100
        backoff_delays = [1, 2, 5, 10, 30, 60]
        rate_limit_delay = 60
        # Cancel previous token refresh task if any
        if self._token_refresh_task and not self._token_refresh_task.done():
            self._token_refresh_task.cancel()
            try:
                await self._token_refresh_task
            except asyncio.CancelledError:
                pass
            self._token_refresh_task = None
        while reconnect_attempts <= max_reconnect_attempts:
            heartbeat_interval = 45000
            should_refresh_token = False
            ws = None
            heartbeat_task = None
            # Refresh token if needed
            if should_refresh_token:
                self.access_token = ''
                self.access_token_expiry_time = None
            try:
                ws_url = await self.get_gateway_url()
                await self.logger.info(f'Gateway URL obtained: {ws_url[:60]}...')
            except Exception as e:
                error_msg = str(e)
                await self.logger.error(f'Failed to get gateway URL: {e}')
                reconnect_attempts += 1
                if '100017' in error_msg or '频率' in error_msg or 'Too many' in error_msg:
                    delay = rate_limit_delay
                else:
                    delay = backoff_delays[min(reconnect_attempts - 1, len(backoff_delays) - 1)]
                await self.logger.info(f'Reconnecting in {delay}s (attempt {reconnect_attempts})')
                await asyncio.sleep(delay)
                continue
            try:
                await self.logger.info('Connecting to WebSocket gateway...')
                ws = await websockets.connect(ws_url)
                await self.logger.info('WebSocket connected')
            except Exception as e:
                await self.logger.error(f'WebSocket connection failed: {e}')
                reconnect_attempts += 1
                delay = backoff_delays[min(reconnect_attempts - 1, len(backoff_delays) - 1)]
                await self.logger.info(f'Reconnecting in {delay}s (attempt {reconnect_attempts})')
                await asyncio.sleep(delay)
                continue
            try:
                async for raw_msg in ws:
                    try:
                        payload = json.loads(raw_msg)
                    except json.JSONDecodeError:
                        await self.logger.error(f'Failed to parse message: {raw_msg}')
                        continue
                    op = payload.get('op')
                    d = payload.get('d', {})
                    s = payload.get('s')
                    t = payload.get('t')
                    if not isinstance(d, dict):
                        d = {}
                    if op == 10:  # Hello
                        heartbeat_interval = d.get('heartbeat_interval', 45000)
                        await self.logger.info(f'Received Hello, heartbeat_interval={heartbeat_interval}ms')
                        # Send Identify or Resume
                        if session_id and last_seq > 0:
                            resume_payload = {
                                'op': 6,
                                'd': {
                                    'token': f'QQBot {self.access_token}',
                                    'session_id': session_id,
                                    'seq': last_seq,
                                },
                            }
                            await ws.send(json.dumps(resume_payload))
                            await self.logger.info(f'Sent Resume, session_id={session_id}, seq={last_seq}')
                        else:
                            identify_payload = {
                                'op': 2,
                                'd': {
                                    'token': f'QQBot {self.access_token}',
                                    'intents': self.FULL_INTENTS,
                                    'shard': [0, 1],
                                },
                            }
                            await ws.send(json.dumps(identify_payload))
                            await self.logger.info(f'Sent Identify, intents={self.FULL_INTENTS}')
                        # Start heartbeat
                        async def _heartbeat_loop(conn, interval_ms):
                            interval_sec = interval_ms / 1000.0
                            try:
                                while True:
                                    await asyncio.sleep(interval_sec)
                                    try:
                                        hb_payload = {'op': 1, 'd': last_seq}
                                        await conn.send(json.dumps(hb_payload))
                                    except Exception:
                                        break
                            except asyncio.CancelledError:
                                pass
                        heartbeat_task = asyncio.create_task(_heartbeat_loop(ws, heartbeat_interval))
                    elif op == 0:  # Dispatch
                        if s is not None:
                            last_seq = s
                        if t == 'READY':
                            session_id = d.get('session_id', '')
                            reconnect_attempts = 0
                            await self.logger.info(f'READY, session_id={session_id}')
                            if on_ready:
                                try:
                                    result = on_ready()
                                    if asyncio.iscoroutine(result):
                                        await result
                                except Exception:
                                    pass
                            # Track token refresh task to avoid leaks
                            if self._token_refresh_task and not self._token_refresh_task.done():
                                self._token_refresh_task.cancel()
                                try:
                                    await self._token_refresh_task
                                except asyncio.CancelledError:
                                    pass
                            self._token_refresh_task = asyncio.create_task(self._background_token_refresh())
                        elif t == 'RESUMED':
                            reconnect_attempts = 0
                            await self.logger.info('RESUMED')
                        else:
                            await self.logger.debug(f'Received event: {t}, seq={s}')
                            if on_event:
                                try:
                                    result = on_event(t, d)
                                    if asyncio.iscoroutine(result):
                                        await result
                                except Exception:
                                    await self.logger.error(f'Error handling event {t}: {traceback.format_exc()}')
                    elif op == 11:  # Heartbeat ACK
                        pass
                    elif op == 7:  # Reconnect
                        await self.logger.info('Received Reconnect directive')
                        break
                    elif op == 9:  # Invalid Session
                        can_resume = d.get('can_resume', False)
                        await self.logger.warning(f'Invalid Session, can_resume={can_resume}')
                        if not can_resume:
                            session_id = ''
                            last_seq = 0
                            should_refresh_token = True
                        break
                # Connection closed normally (end of async for)
                try:
                    close_code = ws.close_code
                    close_reason = ws.close_reason or ''
                except Exception:
                    close_code = None
                    close_reason = ''
                await self.logger.info(f'Connection closed, code={close_code}, reason={close_reason}')
                if close_code == 4004:
                    should_refresh_token = True
                elif close_code in (4006, 4007, 4009):
                    session_id = ''
                    last_seq = 0
                    should_refresh_token = True
                elif close_code == 4008:
                    reconnect_attempts += 1
                    delay = rate_limit_delay
                    await self.logger.info(
                        f'Rate limited, waiting {delay}s before reconnect (attempt {reconnect_attempts})'
                    )
                    await asyncio.sleep(delay)
                    continue
                elif close_code in (4914, 4915):
                    err = Exception(f'Bot disconnected/banned (close_code={close_code})')
                    if on_error:
                        await self._safe_callback(on_error, err)
                    return
                elif close_code in (4900, 4901, 4902, 4903, 4904, 4905, 4906, 4907, 4908, 4909, 4910, 4911, 4912, 4913):
                    session_id = ''
                    last_seq = 0
                if close_code == 1000:
                    return
            except asyncio.CancelledError:
                raise
            except Exception:
                await self.logger.error(f'Unexpected error in WebSocket loop: {traceback.format_exc()}')
            finally:
                if heartbeat_task:
                    heartbeat_task.cancel()
                    try:
                        await heartbeat_task
                    except asyncio.CancelledError:
                        pass
                if ws:
                    try:
                        await ws.close()
                    except Exception:
                        pass
            # If we reach here, we need to reconnect
            reconnect_attempts += 1
            if reconnect_attempts > max_reconnect_attempts:
                await self.logger.error(f'Max reconnect attempts ({max_reconnect_attempts}) reached, stopping')
                if on_error:
                    await self._safe_callback(on_error, Exception('Max reconnect attempts reached'))
                return
            delay = backoff_delays[min(reconnect_attempts - 1, len(backoff_delays) - 1)]
            await self.logger.info(f'Reconnecting in {delay}s (attempt {reconnect_attempts})')
            await asyncio.sleep(delay)
    async def _safe_callback(self, callback, *args):
        """Safely invoke a callback, handling both sync and async functions."""
        try:
            result = callback(*args)
            if asyncio.iscoroutine(result):
                await result
        except Exception:
            pass
    async def connect_gateway_loop(
        self,
        on_event: Callable[[str, dict], Any],
        on_ready: Optional[Callable[[], Any]] = None,
        on_error: Optional[Callable[[Exception], Any]] = None,
    ):
        """持续重连的网关循环。"""
        await self.connect_gateway(on_event, on_ready, on_error)
--- a/src/langbot/pkg/api/http/controller/groups/pipelines/embed.py
+++ b/src/langbot/pkg/api/http/controller/groups/pipelines/embed.py
@@ -0,0 +1,384 @@
 """Embed widget routes - serve embeddable chat widget for external websites.
 All user-facing URLs are keyed by **bot_uuid** (not pipeline_uuid) so that
 internal pipeline identifiers are never exposed to end-users.  Each handler
 resolves the bot_uuid to the owning ``web_page_bot`` RuntimeBot and extracts
 the bound pipeline_uuid for internal routing.
 """
 import asyncio
 import datetime
 import json
 import logging
 import uuid
 import hmac
 import hashlib
 import time
 import re
 import httpx
 import quart
 from ... import group
 from ......utils import paths
 from ......platform.sources.websocket_manager import ws_connection_manager
 logger = logging.getLogger(__name__)
 # Cache the widget template content
 _widget_template_cache: str | None = None
 _logo_bytes_cache: bytes | None = None
 def _is_valid_uuid(s: str) -> bool:
    return bool(re.match(r'^[a-f0-9]{8}-[a-f0-9]{4}-[a-f0-9]{4}-[a-f0-9]{4}-[a-f0-9]{12}$', s))
 def _get_widget_template() -> str:
    """Load and cache the widget JS template."""
    global _widget_template_cache
    if _widget_template_cache is None:
        template_path = paths.get_resource_path('templates/embed/widget.js')
        with open(template_path, 'r', encoding='utf-8') as f:
            _widget_template_cache = f.read()
    return _widget_template_cache
 def _get_logo_bytes() -> bytes:
    """Load and cache the logo image."""
    global _logo_bytes_cache
    if _logo_bytes_cache is None:
        logo_path = paths.get_resource_path('templates/embed/logo.webp')
        with open(logo_path, 'rb') as f:
            _logo_bytes_cache = f.read()
    return _logo_bytes_cache
@group.group_class('embed', '/api/v1/embed')
 class EmbedRouterGroup(group.RouterGroup):
    # -- helpers -------------------------------------------------------------
    def _resolve_bot(self, bot_uuid: str):
        """Resolve *bot_uuid* to ``(runtime_bot, pipeline_uuid)``.
        Returns ``(None, None)`` when the bot does not exist, is not a
        ``web_page_bot``, is disabled, or has no pipeline bound.
        """
        for bot in self.ap.platform_mgr.bots:
            if (
                bot.bot_entity.uuid == bot_uuid
                and bot.bot_entity.adapter == 'web_page_bot'
                and bot.bot_entity.enable
                and bot.bot_entity.use_pipeline_uuid
            ):
                return bot, bot.bot_entity.use_pipeline_uuid
        return None, None
    def _get_bot_config(self, bot_uuid: str) -> dict:
        for bot in self.ap.platform_mgr.bots:
            if bot.bot_entity.uuid == bot_uuid and bot.bot_entity.adapter == 'web_page_bot':
                return bot.bot_entity.adapter_config
        return {}
    async def _verify_session_token(self, request, bot_uuid: str) -> bool:
        config = self._get_bot_config(bot_uuid)
        secret = config.get('turnstile_secret_key', '')
        if not secret:
            return True
        auth_header = request.headers.get('Authorization', '')
        if not auth_header.startswith('Bearer '):
            return False
        token = auth_header[7:]
        try:
            ts_str, mac = token.split('.', 1)
            ts = float(ts_str)
            if time.time() - ts > 86400:
                return False
            expected_mac = hmac.new(secret.encode(), f'{ts_str}'.encode(), hashlib.sha256).hexdigest()
            return hmac.compare_digest(mac, expected_mac)
        except Exception:
            return False
    # -- routes --------------------------------------------------------------
    async def initialize(self) -> None:
        @self.route('/<bot_uuid>/turnstile/verify', methods=['POST'], auth_type=group.AuthType.NONE)
        async def verify_turnstile(bot_uuid: str) -> str:
            if not _is_valid_uuid(bot_uuid):
                return self.http_status(400, -1, 'Invalid bot_uuid format')
            runtime_bot, pipeline_uuid = self._resolve_bot(bot_uuid)
            if runtime_bot is None:
                return self.http_status(404, -1, 'Bot not found or not available')
            try:
                data = await quart.request.get_json()
                token = data.get('token')
                if not token:
                    return self.http_status(400, -1, 'Token is required')
                config = self._get_bot_config(bot_uuid)
                secret = config.get('turnstile_secret_key', '')
                if not secret:
                    ts = time.time()
                    return self.success(data={'token': f'{ts}.dummy'})
                async with httpx.AsyncClient() as client:
                    resp = await client.post(
                        'https://challenges.cloudflare.com/turnstile/v0/siteverify',
                        data={'secret': secret, 'response': token},
                    )
                    result = resp.json()
                if not result.get('success'):
                    return self.http_status(403, -1, 'Turnstile verification failed')
                ts = time.time()
                mac = hmac.new(secret.encode(), f'{ts}'.encode(), hashlib.sha256).hexdigest()
                session_token = f'{ts}.{mac}'
                return self.success(data={'token': session_token})
            except Exception as e:
                logger.error(f'Turnstile verify failed: {e}', exc_info=True)
                return self.http_status(500, -1, 'Internal server error')
        @self.route('/<bot_uuid>/widget.js', methods=['GET'], auth_type=group.AuthType.NONE)
        async def serve_widget(bot_uuid: str) -> quart.Response:
            """Serve the embed widget JavaScript with injected configuration."""
            if not _is_valid_uuid(bot_uuid):
                return self.http_status(400, -1, 'Invalid bot_uuid format')
            runtime_bot, pipeline_uuid = self._resolve_bot(bot_uuid)
            if runtime_bot is None:
                return quart.Response(
                    '// Bot not found or not available', status=404, content_type='application/javascript'
                )
            try:
                template = _get_widget_template()
            except FileNotFoundError:
                return quart.Response('// Widget template not found', status=404, content_type='application/javascript')
            base_url = quart.request.host_url.rstrip('/')
            webhook_prefix = self.ap.instance_config.data.get('api', {}).get('webhook_prefix', '')
            if webhook_prefix:
                base_url = webhook_prefix.rstrip('/')
            if not re.match(r'^https?://[a-zA-Z0-9._:/-]+$', base_url):
                base_url = quart.request.host_url.rstrip('/')
            config = self._get_bot_config(bot_uuid)
            site_key = config.get('turnstile_site_key', '')
            locale = config.get('language', 'en_US') or 'en_US'
            bubble_icon = config.get('bubble_icon', 'logo') or 'logo'
            widget_js = template.replace('__LANGBOT_TURNSTILE_SITE_KEY__', site_key)
            widget_js = widget_js.replace('__LANGBOT_BOT_UUID__', bot_uuid)
            widget_js = widget_js.replace('__LANGBOT_BASE_URL__', base_url)
            widget_js = widget_js.replace('__LANGBOT_LOCALE__', locale)
            widget_js = widget_js.replace('__LANGBOT_BUBBLE_ICON__', bubble_icon)
            response = quart.Response(widget_js, content_type='application/javascript; charset=utf-8')
            response.headers['Cache-Control'] = 'public, max-age=300'
            return response
        @self.route('/logo', methods=['GET'], auth_type=group.AuthType.NONE)
        async def serve_logo() -> quart.Response:
            """Serve the LangBot logo for the embed widget."""
            try:
                logo_data = _get_logo_bytes()
            except FileNotFoundError:
                return quart.Response('', status=404)
            response = quart.Response(logo_data, content_type='image/webp')
            response.headers['Cache-Control'] = 'public, max-age=86400'
            return response
        @self.route('/<bot_uuid>/messages/<session_type>', methods=['GET'], auth_type=group.AuthType.NONE)
        async def get_embed_messages(bot_uuid: str, session_type: str) -> str:
            if not _is_valid_uuid(bot_uuid):
                return self.http_status(400, -1, 'Invalid bot_uuid format')
            runtime_bot, pipeline_uuid = self._resolve_bot(bot_uuid)
            if runtime_bot is None:
                return self.http_status(404, -1, 'Bot not found or not available')
            if not await self._verify_session_token(quart.request, bot_uuid):
                return self.http_status(403, -1, 'Unauthorized or session expired')
            try:
                if session_type not in ['person', 'group']:
                    return self.http_status(400, -1, 'session_type must be person or group')
                websocket_adapter = self.ap.platform_mgr.websocket_proxy_bot.adapter
                if not websocket_adapter:
                    return self.http_status(404, -1, 'WebSocket adapter not found')
                messages = websocket_adapter.get_websocket_messages(pipeline_uuid, session_type)
                return self.success(data={'messages': messages})
            except Exception as e:
                logger.error(f'Failed to get embed messages: {e}', exc_info=True)
                return self.http_status(500, -1, 'Internal server error')
        @self.route('/<bot_uuid>/reset/<session_type>', methods=['POST'], auth_type=group.AuthType.NONE)
        async def reset_embed_session(bot_uuid: str, session_type: str) -> str:
            if not _is_valid_uuid(bot_uuid):
                return self.http_status(400, -1, 'Invalid bot_uuid format')
            runtime_bot, pipeline_uuid = self._resolve_bot(bot_uuid)
            if runtime_bot is None:
                return self.http_status(404, -1, 'Bot not found or not available')
            if not await self._verify_session_token(quart.request, bot_uuid):
                return self.http_status(403, -1, 'Unauthorized or session expired')
            try:
                if session_type not in ['person', 'group']:
                    return self.http_status(400, -1, 'session_type must be person or group')
                websocket_adapter = self.ap.platform_mgr.websocket_proxy_bot.adapter
                if not websocket_adapter:
                    return self.http_status(404, -1, 'WebSocket adapter not found')
                websocket_adapter.reset_session(pipeline_uuid, session_type)
                return self.success(data={'message': 'Session reset successfully'})
            except Exception as e:
                logger.error(f'Failed to reset embed session: {e}', exc_info=True)
                return self.http_status(500, -1, 'Internal server error')
        @self.route('/<bot_uuid>/feedback', methods=['POST'], auth_type=group.AuthType.NONE)
        async def submit_feedback(bot_uuid: str) -> str:
            if not _is_valid_uuid(bot_uuid):
                return self.http_status(400, -1, 'Invalid bot_uuid format')
            runtime_bot, pipeline_uuid = self._resolve_bot(bot_uuid)
            if runtime_bot is None:
                return self.http_status(404, -1, 'Bot not found or not available')
            if not await self._verify_session_token(quart.request, bot_uuid):
                return self.http_status(403, -1, 'Unauthorized or session expired')
            try:
                data = await quart.request.get_json()
                message_id = data.get('message_id', '')
                feedback_type = data.get('feedback_type')
                if feedback_type not in (1, 2, 3):
                    return self.http_status(400, -1, 'feedback_type must be 1 (like), 2 (dislike), or 3 (cancel)')
                feedback_id = f'embed_{uuid.uuid4().hex[:12]}'
                await self.ap.monitoring_service.record_feedback(
                    feedback_id=feedback_id,
                    feedback_type=feedback_type,
                    bot_id=runtime_bot.bot_entity.uuid,
                    bot_name=runtime_bot.bot_entity.name or bot_uuid,
                    pipeline_id=pipeline_uuid,
                    message_id=str(message_id),
                    platform='web_page_bot',
                )
                return self.success(data={'feedback_id': feedback_id})
            except Exception as e:
                logger.error(f'Failed to record feedback: {e}', exc_info=True)
                return self.http_status(500, -1, 'Internal server error')
        # -- Embed WebSocket endpoint ----------------------------------------
        @self.quart_app.websocket(self.path + '/<bot_uuid>/ws/connect')
        async def embed_websocket_connect(bot_uuid: str):
            """WebSocket connection for embed widget, keyed by bot_uuid."""
            if not _is_valid_uuid(bot_uuid):
                await quart.websocket.send(json.dumps({'type': 'error', 'message': 'Invalid bot_uuid format'}))
                return
            runtime_bot, pipeline_uuid = self._resolve_bot(bot_uuid)
            if runtime_bot is None:
                await quart.websocket.send(json.dumps({'type': 'error', 'message': 'Bot not found or not available'}))
                return
            session_type = quart.websocket.args.get('session_type', 'person')
            if session_type not in ['person', 'group']:
                await quart.websocket.send(
                    json.dumps({'type': 'error', 'message': 'session_type must be person or group'})
                )
                return
            websocket_adapter = self.ap.platform_mgr.websocket_proxy_bot.adapter
            if not websocket_adapter:
                await quart.websocket.send(json.dumps({'type': 'error', 'message': 'WebSocket adapter not found'}))
                return
            try:
                connection = await ws_connection_manager.add_connection(
                    websocket=quart.websocket._get_current_object(),
                    pipeline_uuid=pipeline_uuid,
                    session_type=session_type,
                    metadata={'user_agent': quart.websocket.headers.get('User-Agent', '')},
                )
                await quart.websocket.send(
                    json.dumps(
                        {
                            'type': 'connected',
                            'connection_id': connection.connection_id,
                            'bot_uuid': bot_uuid,
                            'session_type': session_type,
                            'timestamp': connection.created_at.isoformat(),
                        }
                    )
                )
                logger.debug(
                    f'Embed WebSocket connected: {connection.connection_id} '
                    f'(bot={bot_uuid}, pipeline={pipeline_uuid}, session_type={session_type})'
                )
                receive_task = asyncio.create_task(self._handle_receive(connection, websocket_adapter, runtime_bot))
                send_task = asyncio.create_task(self._handle_send(connection))
                try:
                    await asyncio.gather(receive_task, send_task)
                except Exception as e:
                    logger.error(f'Embed WebSocket task error: {e}')
                finally:
                    await ws_connection_manager.remove_connection(connection.connection_id)
            except Exception as e:
                logger.error(f'Embed WebSocket connection error: {e}', exc_info=True)
                try:
                    await quart.websocket.send(json.dumps({'type': 'error', 'message': 'Internal server error'}))
                except Exception:
                    pass
    # -- WebSocket receive/send helpers --------------------------------------
    async def _handle_receive(self, connection, websocket_adapter, owner_bot):
        try:
            while connection.is_active:
                message = await quart.websocket.receive()
                await ws_connection_manager.update_activity(connection.connection_id)
                try:
                    data = json.loads(message)
                    message_type = data.get('type', 'message')
                    if message_type == 'ping':
                        await connection.send_queue.put(
                            {'type': 'pong', 'timestamp': datetime.datetime.now().isoformat()}
                        )
                    elif message_type == 'message':
                        await websocket_adapter.handle_websocket_message(connection, data, owner_bot=owner_bot)
                    elif message_type == 'disconnect':
                        break
                except json.JSONDecodeError:
                    await connection.send_queue.put({'type': 'error', 'message': 'Invalid JSON format'})
        except Exception as e:
            logger.error(f'Embed receive error: {e}', exc_info=True)
        finally:
            connection.is_active = False
    async def _handle_send(self, connection):
        try:
            while connection.is_active:
                try:
                    message = await asyncio.wait_for(connection.send_queue.get(), timeout=1.0)
                    await quart.websocket.send(json.dumps(message))
                except asyncio.TimeoutError:
                    continue
        except Exception as e:
            logger.error(f'Embed send error: {e}', exc_info=True)
        finally:
            connection.is_active = False
--- a/src/langbot/pkg/api/http/controller/groups/pipelines/websocket_chat.py
+++ b/src/langbot/pkg/api/http/controller/groups/pipelines/websocket_chat.py
@@ -43,6 +43,9 @@ class WebSocketChatRouterGroup(group.RouterGroup):
                    await quart.websocket.send(json.dumps({'type': 'error', 'message': 'WebSocket adapter not found'}))
                    return
                # Find the owning bot for this pipeline (e.g. a web_page_bot)
                owner_bot = self._find_owner_bot(pipeline_uuid)
                # 注册连接
                connection = await ws_connection_manager.add_connection(
                    websocket=quart.websocket._get_current_object(),
@@ -70,7 +73,7 @@ class WebSocketChatRouterGroup(group.RouterGroup):
                )
                # 创建接收和发送任务
-                receive_task = asyncio.create_task(self._handle_receive(connection, websocket_adapter))
+                receive_task = asyncio.create_task(self._handle_receive(connection, websocket_adapter, owner_bot))
                send_task = asyncio.create_task(self._handle_send(connection))
                # 等待任务完成
@@ -178,7 +181,14 @@ class WebSocketChatRouterGroup(group.RouterGroup):
            except Exception as e:
                return self.http_status(500, -1, f'Internal server error: {str(e)}')
-    async def _handle_receive(self, connection, websocket_adapter):
+    def _find_owner_bot(self, pipeline_uuid: str):
        """Find a user-created bot (e.g. web_page_bot) that owns this pipeline."""
        for bot in self.ap.platform_mgr.bots:
            if bot.bot_entity.adapter == 'web_page_bot' and bot.bot_entity.use_pipeline_uuid == pipeline_uuid:
                return bot
        return None
    async def _handle_receive(self, connection, websocket_adapter, owner_bot=None):
        """处理接收消息的任务"""
        try:
            while connection.is_active:
@@ -203,7 +213,7 @@ class WebSocketChatRouterGroup(group.RouterGroup):
                        logger.debug(f'收到消息: {data} from {connection.connection_id}')
                        # 处理消息（不等待响应，响应会通过broadcast异步发送）
-                        await websocket_adapter.handle_websocket_message(connection, data)
+                        await websocket_adapter.handle_websocket_message(connection, data, owner_bot=owner_bot)
                    elif message_type == 'disconnect':
                        # 客户端主动断开
--- a/src/langbot/pkg/api/http/controller/groups/platform/adapters.py
+++ b/src/langbot/pkg/api/http/controller/groups/platform/adapters.py
@@ -1,5 +1,6 @@
 import quart
 import mimetypes
 import asyncio
 from ... import group
 from langbot.pkg.utils import importutil
@@ -35,3 +36,640 @@ class AdaptersRouterGroup(group.RouterGroup):
            return quart.Response(
                importutil.read_resource_file_bytes(icon_path), mimetype=mimetypes.guess_type(icon_path)[0]
            )
        # In-memory session store for active registrations
        _create_app_sessions: dict = {}
        _SESSION_TTL = 900  # 15 minutes
        def _cleanup_expired_sessions():
            """Remove sessions that have exceeded their TTL."""
            import time
            now = time.time()
            expired = [sid for sid, s in _create_app_sessions.items() if now - s.get('created_at', 0) > _SESSION_TTL]
            for sid in expired:
                session = _create_app_sessions.pop(sid, None)
                if session and session.get('task') and not session['task'].done():
                    session['task'].cancel()
        @self.route('/lark/create-app', methods=['POST'])
        async def _() -> str:
            """Start Feishu one-click app registration. Returns session_id + QR code URL."""
            import uuid
            import time
            import lark_oapi as lark
            from lark_oapi.scene.registration.errors import AppAccessDeniedError, AppExpiredError
            _cleanup_expired_sessions()
            session_id = str(uuid.uuid4())
            loop = asyncio.get_running_loop()
            session = {
                'status': 'pending',
                'qr_url': None,
                'expire_at': None,
                'app_id': None,
                'app_secret': None,
                'error': None,
                'created_at': time.time(),
            }
            _create_app_sessions[session_id] = session
            def on_qr_code(info):
                # May be called from a background thread by the SDK;
                # use call_soon_threadsafe to safely update session state.
                def _update():
                    session['qr_url'] = info['url']
                    session['expire_at'] = time.time() + 600  # 10 minutes
                    session['status'] = 'waiting'
                loop.call_soon_threadsafe(_update)
            async def run_registration():
                try:
                    result = await lark.aregister_app(
                        on_qr_code=on_qr_code,
                        source='langbot',
                    )
                    session['status'] = 'success'
                    session['app_id'] = result['client_id']
                    session['app_secret'] = result['client_secret']
                except AppAccessDeniedError:
                    session['status'] = 'error'
                    session['error'] = 'User denied authorization'
                except AppExpiredError:
                    session['status'] = 'error'
                    session['error'] = 'QR code expired'
                except Exception as e:
                    session['status'] = 'error'
                    session['error'] = str(e)
            task = asyncio.create_task(run_registration())
            session['task'] = task
            # Wait for QR code to be ready (max 10 seconds)
            for _ in range(20):
                if session['qr_url']:
                    break
                await asyncio.sleep(0.5)
            if not session['qr_url']:
                task.cancel()
                session['status'] = 'error'
                session['error'] = 'Timeout waiting for QR code'
                return self.http_status(504, -1, 'Timeout waiting for QR code')
            return self.success(
                data={
                    'session_id': session_id,
                    'qr_url': session['qr_url'],
                    'expire_at': session['expire_at'],
                }
            )
        @self.route('/lark/create-app/status/<session_id>', methods=['GET'])
        async def _(session_id: str) -> str:
            """Poll registration status."""
            session = _create_app_sessions.get(session_id)
            if not session:
                return self.http_status(404, -1, 'Session not found')
            data = {'status': session['status']}
            if session['status'] == 'success':
                data['app_id'] = session['app_id']
                data['app_secret'] = session['app_secret']
                _create_app_sessions.pop(session_id, None)
            elif session['status'] == 'error':
                data['error'] = session['error']
                _create_app_sessions.pop(session_id, None)
            return self.success(data=data)
        @self.route('/lark/create-app/<session_id>', methods=['DELETE'])
        async def _(session_id: str) -> str:
            """Cancel and clean up a registration session."""
            session = _create_app_sessions.pop(session_id, None)
            if session and session.get('task') and not session['task'].done():
                session['task'].cancel()
            return self.success(data={})
        # -----------------------------------------------------------------------
        # WeChat QR Code Login
        # -----------------------------------------------------------------------
        _weixin_login_sessions: dict = {}
        _WEIXIN_SESSION_TTL = 600  # 10 minutes (3 retries × 3 min QR validity)
        def _cleanup_expired_weixin_sessions():
            import time
            now = time.time()
            expired = [
                sid for sid, s in _weixin_login_sessions.items() if now - s.get('created_at', 0) > _WEIXIN_SESSION_TTL
            ]
            for sid in expired:
                session = _weixin_login_sessions.pop(sid, None)
                if session and session.get('task') and not session['task'].done():
                    session['task'].cancel()
        @self.route('/weixin/login', methods=['POST'])
        async def _() -> str:
            """Start WeChat QR code login. Returns session_id + QR code data URL."""
            import uuid
            import time
            import io
            import base64
            from langbot.libs.openclaw_weixin_api.client import OpenClawWeixinClient, DEFAULT_BASE_URL
            _cleanup_expired_weixin_sessions()
            session_id = str(uuid.uuid4())
            loop = asyncio.get_running_loop()
            session = {
                'status': 'pending',
                'qr_data_url': None,
                'expire_at': None,
                'token': None,
                'base_url': None,
                'account_id': None,
                'error': None,
                'created_at': time.time(),
            }
            _weixin_login_sessions[session_id] = session
            client = OpenClawWeixinClient(
                base_url=DEFAULT_BASE_URL,
                token='',
            )
            async def run_login():
                try:
                    import qrcode as qr_lib
                    for _attempt in range(3):
                        qr_resp = await client.fetch_qrcode()
                        if not qr_resp.qrcode or not qr_resp.qrcode_img_content:
                            raise Exception('Failed to get QR code from server')
                        # Generate QR code image locally
                        qr = qr_lib.QRCode(error_correction=qr_lib.constants.ERROR_CORRECT_L)
                        qr.add_data(qr_resp.qrcode_img_content)
                        qr.make(fit=True)
                        img = qr.make_image(fill_color='black', back_color='white')
                        buf = io.BytesIO()
                        img.save(buf, format='PNG')
                        b64 = base64.b64encode(buf.getvalue()).decode('utf-8')
                        data_url = f'data:image/png;base64,{b64}'
                        def _update_qr():
                            session['qr_data_url'] = data_url
                            session['expire_at'] = time.time() + 480  # 8 minutes
                            session['status'] = 'waiting'
                        loop.call_soon_threadsafe(_update_qr)
                        # Poll for scan status
                        deadline = loop.time() + 180
                        while loop.time() < deadline:
                            try:
                                status_resp = await client.poll_qrcode_status(qr_resp.qrcode)
                            except Exception:
                                await asyncio.sleep(2)
                                continue
                            if status_resp.status == 'confirmed' and status_resp.bot_token:
                                session['status'] = 'success'
                                session['token'] = status_resp.bot_token
                                session['base_url'] = status_resp.baseurl or client.base_url
                                session['account_id'] = status_resp.ilink_bot_id or ''
                                return
                            if status_resp.status == 'expired':
                                break  # retry with new QR code
                            await asyncio.sleep(1)
                        else:
                            pass  # timeout, retry
                    # All retries exhausted
                    session['status'] = 'error'
                    session['error'] = 'QR code login failed: max retries exceeded'
                except Exception as e:
                    session['status'] = 'error'
                    session['error'] = str(e)
                finally:
                    await client.close()
            task = asyncio.create_task(run_login())
            session['task'] = task
            # Wait for QR code to be ready (max 10 seconds)
            for _ in range(20):
                if session['qr_data_url']:
                    break
                await asyncio.sleep(0.5)
            if not session['qr_data_url']:
                task.cancel()
                session['status'] = 'error'
                session['error'] = 'Timeout waiting for QR code'
                return self.http_status(504, -1, 'Timeout waiting for QR code')
            return self.success(
                data={
                    'session_id': session_id,
                    'qr_data_url': session['qr_data_url'],
                    'expire_at': session['expire_at'],
                }
            )
        @self.route('/weixin/login/status/<session_id>', methods=['GET'])
        async def _(session_id: str) -> str:
            """Poll WeChat login status."""
            session = _weixin_login_sessions.get(session_id)
            if not session:
                return self.http_status(404, -1, 'Session not found')
            data = {'status': session['status']}
            if session['status'] == 'success':
                data['token'] = session['token']
                data['base_url'] = session['base_url']
                data['account_id'] = session['account_id']
                _weixin_login_sessions.pop(session_id, None)
            elif session['status'] == 'error':
                data['error'] = session['error']
                _weixin_login_sessions.pop(session_id, None)
            return self.success(data=data)
        @self.route('/weixin/login/<session_id>', methods=['DELETE'])
        async def _(session_id: str) -> str:
            """Cancel and clean up a WeChat login session."""
            session = _weixin_login_sessions.pop(session_id, None)
            if session and session.get('task') and not session['task'].done():
                session['task'].cancel()
            return self.success(data={})
        # -----------------------------------------------------------------------
        # DingTalk Device Flow QR Code Login
        # -----------------------------------------------------------------------
        _dingtalk_sessions: dict = {}
        _DINGTALK_SESSION_TTL = 600  # 10 minutes (QR code validity window)
        def _cleanup_expired_dingtalk_sessions():
            import time
            now = time.time()
            expired = [
                sid for sid, s in _dingtalk_sessions.items() if now - s.get('created_at', 0) > _DINGTALK_SESSION_TTL
            ]
            for sid in expired:
                session = _dingtalk_sessions.pop(sid, None)
                if session and session.get('task') and not session['task'].done():
                    session['task'].cancel()
        @self.route('/dingtalk/create-app', methods=['POST'])
        async def _() -> str:
            """Start DingTalk one-click app creation via Device Flow. Returns session_id + QR code URL."""
            import uuid
            import time
            import aiohttp
            DINGTALK_BASE_URL = 'https://oapi.dingtalk.com'
            _cleanup_expired_dingtalk_sessions()
            session_id = str(uuid.uuid4())
            session = {
                'status': 'pending',
                'qr_url': None,
                'expire_at': None,
                'client_id': None,
                'client_secret': None,
                'error': None,
                'created_at': time.time(),
                'device_code': None,
                'interval': 5,
            }
            _dingtalk_sessions[session_id] = session
            async def run_device_flow():
                try:
                    timeout = aiohttp.ClientTimeout(total=10)
                    async with aiohttp.ClientSession(timeout=timeout) as http:
                        # Step 1: Init — get nonce
                        async with http.post(
                            f'{DINGTALK_BASE_URL}/app/registration/init',
                            json={'source': 'langbot'},
                        ) as resp:
                            try:
                                data = await resp.json()
                            except (aiohttp.ContentTypeError, ValueError):
                                session['status'] = 'error'
                                session['error'] = 'Invalid response from DingTalk service'
                                return
                            if data.get('errcode', -1) != 0:
                                session['status'] = 'error'
                                session['error'] = data.get('errmsg', 'Failed to init')
                                return
                            nonce = data['nonce']
                        # Step 2: Begin — get device_code + QR URL
                        async with http.post(
                            f'{DINGTALK_BASE_URL}/app/registration/begin',
                            json={'nonce': nonce},
                        ) as resp:
                            try:
                                data = await resp.json()
                            except (aiohttp.ContentTypeError, ValueError):
                                session['status'] = 'error'
                                session['error'] = 'Invalid response from DingTalk service'
                                return
                            if data.get('errcode', -1) != 0:
                                session['status'] = 'error'
                                session['error'] = data.get('errmsg', 'Failed to begin authorization')
                                return
                            device_code = data['device_code']
                            verification_uri_complete = data.get('verification_uri_complete', '')
                            expires_in = data.get('expires_in', 7200)
                            interval = data.get('interval', 5)
                            session['device_code'] = device_code
                            session['interval'] = interval
                            session['qr_url'] = verification_uri_complete
                            session['expire_at'] = time.time() + 600  # QR code valid for ~10 min
                            session['status'] = 'waiting'
                        # Step 3: Poll for authorization result
                        deadline = time.time() + expires_in
                        while time.time() < deadline:
                            await asyncio.sleep(interval)
                            async with http.post(
                                f'{DINGTALK_BASE_URL}/app/registration/poll',
                                json={'device_code': device_code},
                            ) as poll_resp:
                                try:
                                    poll_data = await poll_resp.json()
                                except (aiohttp.ContentTypeError, ValueError):
                                    continue
                                if poll_data.get('errcode', -1) != 0:
                                    session['status'] = 'error'
                                    session['error'] = poll_data.get('errmsg', 'Poll failed')
                                    return
                                status = poll_data.get('status', '')
                                if status == 'SUCCESS':
                                    session['status'] = 'success'
                                    session['client_id'] = poll_data.get('client_id', '')
                                    session['client_secret'] = poll_data.get('client_secret', '')
                                    return
                                elif status == 'FAIL':
                                    session['status'] = 'error'
                                    session['error'] = poll_data.get('fail_reason', 'Authorization failed')
                                    return
                                elif status == 'EXPIRED':
                                    session['status'] = 'error'
                                    session['error'] = 'QR code expired'
                                    return
                                # status == 'WAITING': continue polling
                        # Timeout
                        session['status'] = 'error'
                        session['error'] = 'QR code expired'
                except asyncio.CancelledError:
                    return
                except Exception as e:
                    session['status'] = 'error'
                    session['error'] = str(e)
            task = asyncio.create_task(run_device_flow())
            session['task'] = task
            # Wait for QR code to be ready (max 10 seconds)
            for _ in range(20):
                if session['qr_url'] or session['error']:
                    break
                await asyncio.sleep(0.5)
            if session['error']:
                task.cancel()
                return self.http_status(502, -1, session['error'])
            if not session['qr_url']:
                task.cancel()
                session['status'] = 'error'
                session['error'] = 'Timeout waiting for QR code'
                return self.http_status(504, -1, 'Timeout waiting for QR code')
            return self.success(
                data={
                    'session_id': session_id,
                    'qr_url': session['qr_url'],
                    'expire_at': session['expire_at'],
                }
            )
        @self.route('/dingtalk/create-app/status/<session_id>', methods=['GET'])
        async def _(session_id: str) -> str:
            """Poll DingTalk Device Flow status."""
            _cleanup_expired_dingtalk_sessions()
            session = _dingtalk_sessions.get(session_id)
            if not session:
                return self.http_status(404, -1, 'Session not found')
            data = {'status': session['status']}
            if session['status'] == 'success':
                data['client_id'] = session['client_id']
                data['client_secret'] = session['client_secret']
                _dingtalk_sessions.pop(session_id, None)
            elif session['status'] == 'error':
                data['error'] = session['error']
                _dingtalk_sessions.pop(session_id, None)
            return self.success(data=data)
        @self.route('/dingtalk/create-app/<session_id>', methods=['DELETE'])
        async def _(session_id: str) -> str:
            """Cancel and clean up a DingTalk Device Flow session."""
            session = _dingtalk_sessions.pop(session_id, None)
            if session and session.get('task') and not session['task'].done():
                session['task'].cancel()
            return self.success(data={})
        # -----------------------------------------------------------------------
        # WeComBot QR Code One-Click Create
        # -----------------------------------------------------------------------
        _wecombot_sessions: dict = {}
        _WECOMBOT_SESSION_TTL = 300  # 5 minutes (WeCom QR validity window)
        def _cleanup_expired_wecombot_sessions():
            import time
            now = time.time()
            expired = [
                sid for sid, s in _wecombot_sessions.items() if now - s.get('created_at', 0) > _WECOMBOT_SESSION_TTL
            ]
            for sid in expired:
                session = _wecombot_sessions.pop(sid, None)
                if session and session.get('task') and not session['task'].done():
                    session['task'].cancel()
        @self.route('/wecombot/create-bot', methods=['POST'])
        async def _() -> str:
            """Start WeComBot one-click creation via QR code. Returns session_id + QR code URL."""
            import uuid
            import time
            import aiohttp
            WECOM_QC_GENERATE_URL = 'https://work.weixin.qq.com/ai/qc/generate'
            WECOM_QC_QUERY_URL = 'https://work.weixin.qq.com/ai/qc/query_result'
            _cleanup_expired_wecombot_sessions()
            session_id = str(uuid.uuid4())
            session = {
                'status': 'pending',
                'qr_url': None,
                'expire_at': None,
                'botid': None,
                'secret': None,
                'error': None,
                'created_at': time.time(),
                'scode': None,
                'task': None,
            }
            _wecombot_sessions[session_id] = session
            async def run_qr_flow():
                try:
                    timeout = aiohttp.ClientTimeout(total=10)
                    async with aiohttp.ClientSession(timeout=timeout) as http:
                        # Step 1: Generate QR code
                        async with http.get(
                            f'{WECOM_QC_GENERATE_URL}?source=langbot&plat=0',
                        ) as resp:
                            try:
                                data = await resp.json()
                            except (aiohttp.ContentTypeError, ValueError):
                                session['status'] = 'error'
                                session['error'] = 'Invalid response from WeCom service'
                                return
                            if not data.get('data', {}).get('scode') or not data.get('data', {}).get('auth_url'):
                                session['status'] = 'error'
                                session['error'] = data.get('errmsg', 'Failed to generate QR code')
                                return
                            scode = data['data']['scode']
                            auth_url = data['data']['auth_url']
                            session['scode'] = scode
                            session['qr_url'] = auth_url
                            session['expire_at'] = time.time() + _WECOMBOT_SESSION_TTL
                            session['status'] = 'waiting'
                        # Step 2: Poll for scan result
                        deadline = time.time() + _WECOMBOT_SESSION_TTL
                        while time.time() < deadline:
                            await asyncio.sleep(3)
                            async with http.get(
                                f'{WECOM_QC_QUERY_URL}?scode={scode}',
                            ) as poll_resp:
                                try:
                                    poll_data = await poll_resp.json()
                                except (aiohttp.ContentTypeError, ValueError):
                                    continue
                                status = poll_data.get('data', {}).get('status', '')
                                if status == 'success':
                                    bot_info = poll_data.get('data', {}).get('bot_info', {})
                                    if bot_info.get('botid') and bot_info.get('secret'):
                                        session['status'] = 'success'
                                        session['botid'] = bot_info['botid']
                                        session['secret'] = bot_info['secret']
                                        return
                                    else:
                                        session['status'] = 'error'
                                        session['error'] = 'Scan succeeded but bot info is incomplete'
                                        return
                        # Timeout
                        session['status'] = 'error'
                        session['error'] = 'QR code expired'
                except asyncio.CancelledError:
                    return
                except Exception as e:
                    session['status'] = 'error'
                    session['error'] = str(e)
            task = asyncio.create_task(run_qr_flow())
            session['task'] = task
            # Wait for QR code to be ready (max 10 seconds)
            for _ in range(20):
                if session['qr_url'] or session['error']:
                    break
                await asyncio.sleep(0.5)
            if session['error']:
                task.cancel()
                return self.http_status(502, -1, session['error'])
            if not session['qr_url']:
                task.cancel()
                session['status'] = 'error'
                session['error'] = 'Timeout waiting for QR code'
                return self.http_status(504, -1, 'Timeout waiting for QR code')
            return self.success(
                data={
                    'session_id': session_id,
                    'qr_url': session['qr_url'],
                    'expire_at': session['expire_at'],
                }
            )
        @self.route('/wecombot/create-bot/status/<session_id>', methods=['GET'])
        async def _(session_id: str) -> str:
            """Poll WeComBot creation status."""
            _cleanup_expired_wecombot_sessions()
            session = _wecombot_sessions.get(session_id)
            if not session:
                return self.http_status(404, -1, 'Session not found')
            data = {'status': session['status']}
            if session['status'] == 'success':
                data['botid'] = session['botid']
                data['secret'] = session['secret']
                _wecombot_sessions.pop(session_id, None)
            elif session['status'] == 'error':
                data['error'] = session['error']
                _wecombot_sessions.pop(session_id, None)
            return self.success(data=data)
        @self.route('/wecombot/create-bot/<session_id>', methods=['DELETE'])
        async def _(session_id: str) -> str:
            """Cancel and clean up a WeComBot creation session."""
            session = _wecombot_sessions.pop(session_id, None)
            if session and session.get('task') and not session['task'].done():
                session['task'].cancel()
            return self.success(data={})
--- a/src/langbot/pkg/api/http/controller/groups/plugins.py
+++ b/src/langbot/pkg/api/http/controller/groups/plugins.py
@@ -6,11 +6,48 @@ import re
 import httpx
 import uuid
 import os
 import posixpath
 from .....core import taskmgr
 from .. import group
 from langbot_plugin.runtime.plugin.mgr import PluginInstallSource
 # Resolve the built-in page SDK JS from the langbot_plugin package
 _PAGE_SDK_PATH = None
 try:
    import langbot_plugin.assets as _assets_pkg
    _candidate = os.path.join(os.path.dirname(_assets_pkg.__file__), 'langbot-page-sdk.js')
    if os.path.exists(_candidate):
        _PAGE_SDK_PATH = _candidate
 except Exception:
    pass
 def _normalize_plugin_asset_path(filepath: str) -> str | None:
    filepath = filepath.replace('\\', '/')
    if filepath.startswith('/'):
        return None
    normalized = posixpath.normpath(filepath)
    if normalized == '.' or normalized.startswith('../') or normalized == '..':
        return None
    if normalized.startswith('components/pages/'):
        return normalized
    return f'assets/{normalized}'
 def _get_request_origin() -> str:
    """Return the public request origin, respecting reverse-proxy headers."""
    forwarded_proto = quart.request.headers.get('X-Forwarded-Proto', '').split(',')[0].strip()
    forwarded_host = quart.request.headers.get('X-Forwarded-Host', '').split(',')[0].strip()
    scheme = forwarded_proto or quart.request.scheme
    host = forwarded_host or quart.request.host
    return f'{scheme}://{host}'
@group.group_class('plugins', '/api/v1/plugins')
 class PluginsRouterGroup(group.RouterGroup):
@@ -27,6 +64,15 @@ class PluginsRouterGroup(group.RouterGroup):
        return None
    async def initialize(self) -> None:
        @self.route('/_sdk/page-sdk.js', methods=['GET'], auth_type=group.AuthType.NONE)
        async def _() -> quart.Response:
            """Serve the built-in LangBot page SDK JavaScript."""
            if _PAGE_SDK_PATH and os.path.exists(_PAGE_SDK_PATH):
                with open(_PAGE_SDK_PATH, 'r') as f:
                    content = f.read()
                return quart.Response(content, mimetype='application/javascript')
            return quart.Response('// SDK not found', status=404, mimetype='application/javascript')
        @self.route('', methods=['GET'], auth_type=group.AuthType.USER_TOKEN_OR_API_KEY)
        async def _() -> str:
            plugins = await self.ap.plugin_connector.list_plugins()
@@ -135,15 +181,62 @@ class PluginsRouterGroup(group.RouterGroup):
            return quart.Response(icon_data, mimetype=mime_type)
        @self.route(
-            '/<author>/<plugin_name>/assets/<filepath>',
+            '/<author>/<plugin_name>/assets/<path:filepath>',
            methods=['GET'],
            auth_type=group.AuthType.NONE,
        )
        async def _(author: str, plugin_name: str, filepath: str) -> quart.Response:
-            asset_data = await self.ap.plugin_connector.get_plugin_assets(author, plugin_name, filepath)
+            asset_path = _normalize_plugin_asset_path(filepath)
            if asset_path is None:
                return quart.Response('Asset not found', status=404)
            asset_data = await self.ap.plugin_connector.get_plugin_assets(author, plugin_name, asset_path)
            if not asset_data.get('asset_base64'):
                return quart.Response('Asset not found', status=404)
            asset_bytes = base64.b64decode(asset_data['asset_base64'])
            mime_type = asset_data['mime_type']
-            return quart.Response(asset_bytes, mimetype=mime_type)
+            resp = quart.Response(asset_bytes, mimetype=mime_type)
            # CSP for HTML pages served to sandboxed iframes (opaque origin).
            # 'self' doesn't work in sandboxed iframes — use actual server origin.
            if mime_type and mime_type.startswith('text/html'):
                origin = _get_request_origin()
                resp.headers['Content-Security-Policy'] = (
                    f'default-src {origin}; '
                    f"script-src {origin} 'unsafe-inline'; "
                    f"style-src {origin} 'unsafe-inline'; "
                    f'img-src {origin} data:; '
                    f'connect-src {origin}; '
                    "frame-src 'none'; "
                    "object-src 'none'"
                )
            return resp
        @self.route(
            '/<author>/<plugin_name>/page-api',
            methods=['POST'],
            auth_type=group.AuthType.USER_TOKEN_OR_API_KEY,
        )
        async def _(author: str, plugin_name: str) -> str:
            """Forward a page API request to the plugin."""
            data = await quart.request.json
            if not isinstance(data, dict):
                return self.http_status(400, -1, 'invalid request body')
            page_id = data.get('page_id', '')
            endpoint = data.get('endpoint', '')
            method = data.get('method', 'POST')
            body = data.get('body')
            if not isinstance(page_id, str) or not isinstance(endpoint, str) or not isinstance(method, str):
                return self.http_status(400, -1, 'invalid page api request')
            if not endpoint.startswith('/') or '..' in endpoint:
                return self.http_status(400, -1, 'invalid endpoint')
            result = await self.ap.plugin_connector.handle_page_api(
                author, plugin_name, page_id, endpoint, method.upper(), body
            )
            if result.get('error'):
                return self.http_status(400, -1, result['error'])
            return self.success(data=result.get('data'))
        @self.route('/github/releases', methods=['POST'], auth_type=group.AuthType.USER_TOKEN_OR_API_KEY)
        async def _() -> str:
--- a/src/langbot/pkg/api/http/controller/groups/system.py
+++ b/src/langbot/pkg/api/http/controller/groups/system.py
@@ -136,6 +136,10 @@ class SystemRouterGroup(group.RouterGroup):
            return self.success(data=task.to_dict())
        @self.route('/storage-analysis', methods=['GET'], auth_type=group.AuthType.USER_TOKEN)
        async def _() -> str:
            return self.success(data=await self.ap.maintenance_service.get_storage_analysis())
        @self.route('/debug/exec', methods=['POST'], auth_type=group.AuthType.USER_TOKEN)
        async def _() -> str:
            if not constants.debug_mode:
--- a/src/langbot/pkg/api/http/controller/groups/user.py
+++ b/src/langbot/pkg/api/http/controller/groups/user.py
@@ -146,6 +146,7 @@ class UserRouterGroup(group.RouterGroup):
                return self.fail(3, str(e))
            except ValueError as e:
                traceback.print_exc()
                self.ap.logger.warning(f'Space OAuth callback failed: {e}')
                return self.fail(1, str(e))
            except Exception as e:
                traceback.print_exc()
--- a/src/langbot/pkg/api/http/service/bot.py
+++ b/src/langbot/pkg/api/http/service/bot.py
@@ -99,11 +99,11 @@ class BotService:
        # TODO: 检查配置信息格式
        bot_data['uuid'] = str(uuid.uuid4())
-        # checkout the default pipeline
+        # bind the most recently updated pipeline if any exist
        result = await self.ap.persistence_mgr.execute_async(
-            sqlalchemy.select(persistence_pipeline.LegacyPipeline).where(
+            sqlalchemy.select(persistence_pipeline.LegacyPipeline)
-                persistence_pipeline.LegacyPipeline.is_default == True
+            .order_by(persistence_pipeline.LegacyPipeline.updated_at.desc())
-            )
+            .limit(1)
        )
        pipeline = result.first()
        if pipeline is not None:
@@ -120,24 +120,26 @@ class BotService:
    async def update_bot(self, bot_uuid: str, bot_data: dict) -> None:
        """Update bot"""
-        if 'uuid' in bot_data:
+        update_data = bot_data.copy()
-            del bot_data['uuid']
+
        if 'uuid' in update_data:
            del update_data['uuid']
        # set use_pipeline_name
-        if 'use_pipeline_uuid' in bot_data:
+        if 'use_pipeline_uuid' in update_data:
            result = await self.ap.persistence_mgr.execute_async(
                sqlalchemy.select(persistence_pipeline.LegacyPipeline).where(
-                    persistence_pipeline.LegacyPipeline.uuid == bot_data['use_pipeline_uuid']
+                    persistence_pipeline.LegacyPipeline.uuid == update_data['use_pipeline_uuid']
                )
            )
            pipeline = result.first()
            if pipeline is not None:
-                bot_data['use_pipeline_name'] = pipeline.name
+                update_data['use_pipeline_name'] = pipeline.name
            else:
                raise Exception('Pipeline not found')
        await self.ap.persistence_mgr.execute_async(
-            sqlalchemy.update(persistence_bot.Bot).values(bot_data).where(persistence_bot.Bot.uuid == bot_uuid)
+            sqlalchemy.update(persistence_bot.Bot).values(update_data).where(persistence_bot.Bot.uuid == bot_uuid)
        )
        await self.ap.platform_mgr.remove_bot(bot_uuid)
--- a/src/langbot/pkg/api/http/service/knowledge.py
+++ b/src/langbot/pkg/api/http/service/knowledge.py
@@ -31,15 +31,126 @@ class KnowledgeService:
        if not knowledge_engine_plugin_id:
            raise ValueError('knowledge_engine_plugin_id is required')
        creation_settings = kb_data.get('creation_settings', {})
        retrieval_settings = kb_data.get('retrieval_settings', {})
        # Validate required fields based on plugin's creation_schema and retrieval_schema
        await self._validate_schema_required_fields(
            knowledge_engine_plugin_id,
            creation_settings,
            retrieval_settings,
        )
        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', {}),
+            creation_settings=creation_settings,
-            retrieval_settings=kb_data.get('retrieval_settings', {}),
+            retrieval_settings=retrieval_settings,
            description=kb_data.get('description', ''),
        )
        return kb.uuid
    async def _validate_schema_required_fields(
        self,
        plugin_id: str,
        creation_settings: dict,
        retrieval_settings: dict,
    ) -> None:
        """Validate required fields based on plugin's creation_schema and retrieval_schema.
        This is a business-agnostic validation that checks all fields marked as
        required in the plugin's schema, regardless of field type.
        Args:
            plugin_id: Knowledge Engine plugin ID.
            creation_settings: User-provided creation settings.
            retrieval_settings: User-provided retrieval settings.
        Raises:
            ValueError: If any required field is missing or empty.
        """
        # Validate creation_schema
        try:
            creation_schema = await self.ap.plugin_connector.get_rag_creation_schema(plugin_id)
            self._check_required_fields(creation_schema, creation_settings, 'creation_settings')
        except ValueError:
            raise
        except Exception as e:
            self.ap.logger.warning(f'Failed to get creation_schema for validation: {e}')
        # Validate retrieval_schema
        try:
            retrieval_schema = await self.ap.plugin_connector.get_rag_retrieval_schema(plugin_id)
            self._check_required_fields(retrieval_schema, retrieval_settings, 'retrieval_settings')
        except ValueError:
            raise
        except Exception as e:
            self.ap.logger.warning(f'Failed to get retrieval_schema for validation: {e}')
    def _check_required_fields(
        self,
        schema: dict | list,
        settings: dict,
        context: str,
    ) -> None:
        """Check required fields in schema against provided settings.
        Args:
            schema: Plugin-defined schema (can be list or dict with 'schema' key).
            settings: User-provided settings values.
            context: Context name for error messages (e.g., 'creation_settings').
        Raises:
            ValueError: If a required field is missing or empty.
        """
        if not schema:
            return
        # schema can be a list directly, or a dict with 'schema' key
        items = schema if isinstance(schema, list) else schema.get('schema', [])
        if not items:
            return
        for item in items:
            field_name = item.get('name')
            if not field_name:
                continue
            is_required = item.get('required', False)
            if not is_required:
                continue
            # Check show_if condition - if field is conditionally shown, only validate when condition is met
            show_if = item.get('show_if')
            if show_if:
                depend_field = show_if.get('field')
                operator = show_if.get('operator')
                expected_value = show_if.get('value')
                if depend_field and operator:
                    depend_value = settings.get(depend_field)
                    # If show_if condition is not met, skip validation for this field
                    if operator == 'eq' and depend_value != expected_value:
                        continue
                    if operator == 'neq' and depend_value == expected_value:
                        continue
                    if operator == 'in' and isinstance(expected_value, list) and depend_value not in expected_value:
                        continue
            value = settings.get(field_name)
            # Validate required field has a non-empty value
            if value is None or (isinstance(value, str) and value.strip() == ''):
                # Get field label for friendly error message
                label = item.get('label', {})
                field_label = (
                    label.get('en_US', field_name)
                    or label.get('zh_Hans', field_name)
                    or label.get('zh_Hant', field_name)
                    or field_name
                )
                raise ValueError(f'{field_label} is required ({context}.{field_name})')
    async def update_knowledge_base(self, kb_uuid: str, kb_data: dict) -> None:
        """更新知识库"""
        # Filter to only mutable fields
--- a/src/langbot/pkg/api/http/service/maintenance.py
+++ b/src/langbot/pkg/api/http/service/maintenance.py
@@ -0,0 +1,309 @@
 from __future__ import annotations
 import datetime
 import os
 import re
 from pathlib import Path
 from typing import Any
 import sqlalchemy
 from ....core import app
 from ....entity.persistence import bstorage as persistence_bstorage
 from ....entity.persistence import monitoring as persistence_monitoring
 LOG_FILE_PATTERN = re.compile(r'^langbot-(\d{4}-\d{2}-\d{2})\.log(?:\.\d+)?$')
 DEFAULT_UPLOAD_FILE_RETENTION_DAYS = 7
 DEFAULT_LOG_RETENTION_DAYS = 3
 class MaintenanceService:
    """Storage maintenance and diagnostics."""
    ap: app.Application
    def __init__(self, ap: app.Application) -> None:
        self.ap = ap
    async def cleanup_expired_files(self) -> dict[str, int]:
        cleanup_cfg = self.ap.instance_config.data.get('storage', {}).get('cleanup', {})
        upload_retention_days = self._positive_int(
            cleanup_cfg.get('uploaded_file_retention_days'),
            DEFAULT_UPLOAD_FILE_RETENTION_DAYS,
            'storage.cleanup.uploaded_file_retention_days',
        )
        log_retention_days = self._positive_int(
            cleanup_cfg.get('log_retention_days'),
            DEFAULT_LOG_RETENTION_DAYS,
            'storage.cleanup.log_retention_days',
        )
        return {
            'uploaded_files': await self._cleanup_expired_uploaded_files(upload_retention_days),
            'log_files': self._cleanup_expired_log_files(log_retention_days),
        }
    async def get_storage_analysis(self) -> dict[str, Any]:
        cleanup_cfg = self.ap.instance_config.data.get('storage', {}).get('cleanup', {})
        upload_retention_days = self._positive_int(
            cleanup_cfg.get('uploaded_file_retention_days'),
            DEFAULT_UPLOAD_FILE_RETENTION_DAYS,
            'storage.cleanup.uploaded_file_retention_days',
        )
        log_retention_days = self._positive_int(
            cleanup_cfg.get('log_retention_days'),
            DEFAULT_LOG_RETENTION_DAYS,
            'storage.cleanup.log_retention_days',
        )
        database_cfg = self.ap.instance_config.data.get('database', {})
        database_type = database_cfg.get('use', 'sqlite')
        database_path = (
            Path(database_cfg.get('sqlite', {}).get('path', 'data/langbot.db')) if database_type == 'sqlite' else None
        )
        roots: list[tuple[str, Path | None]] = [
            ('database', database_path),
            ('logs', Path('data/logs')),
            ('storage', Path('data/storage')),
            ('vector_store', Path('data/chroma')),
            ('plugins', Path('data/plugins')),
            ('mcp', Path('data/mcp')),
            ('temp', Path('data/temp')),
        ]
        sections = []
        for key, path in roots:
            sections.append(
                {
                    'key': key,
                    'path': str(path) if path else '',
                    'exists': path.exists() if path else False,
                    'size_bytes': self._path_size(path) if path else 0,
                    'file_count': self._file_count(path) if path else 0,
                }
            )
        monitoring_counts = await self._monitoring_counts()
        binary_storage = await self._binary_storage_stats()
        upload_candidates = await self._expired_uploaded_candidates(upload_retention_days)
        log_candidates = self._expired_log_candidates(log_retention_days)
        return {
            'generated_at': datetime.datetime.now(datetime.timezone.utc).isoformat(),
            'cleanup_policy': {
                'uploaded_file_retention_days': upload_retention_days,
                'log_retention_days': log_retention_days,
            },
            'sections': sections,
            'database': {
                'type': database_type,
                'monitoring_counts': monitoring_counts,
                'binary_storage': binary_storage,
            },
            'cleanup_candidates': {
                'uploaded_files': upload_candidates,
                'log_files': log_candidates,
            },
            'tasks': self.ap.task_mgr.get_stats() if self.ap.task_mgr else {},
        }
    async def _cleanup_expired_uploaded_files(self, retention_days: int) -> int:
        provider = self.ap.storage_mgr.storage_provider
        provider_name = provider.__class__.__name__
        if provider_name == 'LocalStorageProvider':
            candidates = self._expired_local_upload_candidates(retention_days, include_paths=True)
            deleted = 0
            for item in candidates:
                try:
                    os.remove(item['path'])
                    deleted += 1
                except FileNotFoundError:
                    pass
                except Exception as e:
                    self.ap.logger.warning(f'Failed to delete expired uploaded file {item["key"]}: {e}')
            return deleted
        if provider_name == 'S3StorageProvider':
            return await self._cleanup_expired_s3_uploaded_files(retention_days)
        return 0
    async def _expired_uploaded_candidates(self, retention_days: int) -> list[dict[str, Any]]:
        provider_name = self.ap.storage_mgr.storage_provider.__class__.__name__
        if provider_name == 'LocalStorageProvider':
            return self._expired_local_upload_candidates(retention_days)
        if provider_name == 'S3StorageProvider':
            return await self._expired_s3_upload_candidates(retention_days)
        return []
    async def _cleanup_expired_s3_uploaded_files(self, retention_days: int) -> int:
        provider = self.ap.storage_mgr.storage_provider
        candidates = await self._expired_s3_upload_candidates(retention_days)
        deleted = 0
        for item in candidates:
            await provider.delete(item['key'])
            deleted += 1
        return deleted
    async def _expired_s3_upload_candidates(self, retention_days: int) -> list[dict[str, Any]]:
        provider = self.ap.storage_mgr.storage_provider
        cutoff = datetime.datetime.now(datetime.timezone.utc) - datetime.timedelta(days=retention_days)
        candidates = []
        paginator = provider.s3_client.get_paginator('list_objects_v2')
        for page in paginator.paginate(Bucket=provider.bucket_name):
            for obj in page.get('Contents', []):
                key = obj.get('Key', '')
                last_modified = obj.get('LastModified')
                if not self._is_uploaded_file_key(key):
                    continue
                if last_modified and last_modified < cutoff:
                    candidates.append(
                        {
                            'key': key,
                            'size_bytes': obj.get('Size', 0),
                            'modified_at': last_modified.isoformat(),
                        }
                    )
        return candidates
    def _cleanup_expired_log_files(self, retention_days: int) -> int:
        deleted = 0
        for item in self._expired_log_candidates(retention_days, include_paths=True):
            try:
                os.remove(item['path'])
                deleted += 1
            except FileNotFoundError:
                pass
            except Exception as e:
                self.ap.logger.warning(f'Failed to delete expired log file {item["name"]}: {e}')
        return deleted
    def _expired_local_upload_candidates(
        self, retention_days: int, include_paths: bool = False
    ) -> list[dict[str, Any]]:
        storage_root = Path('data/storage')
        if not storage_root.exists():
            return []
        cutoff = datetime.datetime.now().timestamp() - retention_days * 86400
        candidates = []
        for entry in storage_root.iterdir():
            if not entry.is_file() or not self._is_uploaded_file_key(entry.name):
                continue
            stat = entry.stat()
            if stat.st_mtime >= cutoff:
                continue
            item = {
                'key': entry.name,
                'size_bytes': stat.st_size,
                'modified_at': datetime.datetime.fromtimestamp(stat.st_mtime, datetime.timezone.utc).isoformat(),
            }
            if include_paths:
                item['path'] = str(entry)
            candidates.append(item)
        return candidates
    def _expired_log_candidates(self, retention_days: int, include_paths: bool = False) -> list[dict[str, Any]]:
        log_root = Path('data/logs')
        if not log_root.exists():
            return []
        cutoff_date = datetime.date.today() - datetime.timedelta(days=retention_days - 1)
        candidates = []
        for entry in log_root.iterdir():
            if not entry.is_file():
                continue
            match = LOG_FILE_PATTERN.match(entry.name)
            if not match:
                continue
            try:
                file_date = datetime.date.fromisoformat(match.group(1))
            except ValueError:
                continue
            if file_date >= cutoff_date:
                continue
            stat = entry.stat()
            item = {
                'name': entry.name,
                'date': file_date.isoformat(),
                'size_bytes': stat.st_size,
            }
            if include_paths:
                item['path'] = str(entry)
            candidates.append(item)
        return candidates
    def _is_uploaded_file_key(self, key: str) -> bool:
        return '/' not in key and not key.startswith('plugin_config_')
    async def _monitoring_counts(self) -> dict[str, int]:
        tables = {
            'messages': persistence_monitoring.MonitoringMessage.id,
            'llm_calls': persistence_monitoring.MonitoringLLMCall.id,
            'embedding_calls': persistence_monitoring.MonitoringEmbeddingCall.id,
            'errors': persistence_monitoring.MonitoringError.id,
            'sessions': persistence_monitoring.MonitoringSession.session_id,
            'feedback': persistence_monitoring.MonitoringFeedback.id,
        }
        counts: dict[str, int] = {}
        for key, column in tables.items():
            result = await self.ap.persistence_mgr.execute_async(sqlalchemy.select(sqlalchemy.func.count(column)))
            counts[key] = result.scalar() or 0
        return counts
    async def _binary_storage_stats(self) -> dict[str, Any]:
        count_result = await self.ap.persistence_mgr.execute_async(
            sqlalchemy.select(sqlalchemy.func.count(persistence_bstorage.BinaryStorage.unique_key))
        )
        size_bytes = None
        try:
            size_result = await self.ap.persistence_mgr.execute_async(
                sqlalchemy.select(sqlalchemy.func.sum(sqlalchemy.func.length(persistence_bstorage.BinaryStorage.value)))
            )
            size_bytes = size_result.scalar() or 0
        except Exception as e:
            self.ap.logger.warning(f'Failed to estimate binary storage size: {e}')
        return {
            'count': count_result.scalar() or 0,
            'size_bytes': size_bytes,
        }
    def _path_size(self, path: Path) -> int:
        if not path.exists():
            return 0
        if path.is_file():
            return path.stat().st_size
        total = 0
        for root, _, files in os.walk(path):
            for file_name in files:
                file_path = Path(root) / file_name
                try:
                    total += file_path.stat().st_size
                except FileNotFoundError:
                    pass
        return total
    def _file_count(self, path: Path) -> int:
        if not path.exists():
            return 0
        if path.is_file():
            return 1
        count = 0
        for _, _, files in os.walk(path):
            count += len(files)
        return count
    def _positive_int(self, value: Any, default: int, name: str) -> int:
        try:
            parsed = int(value)
        except (TypeError, ValueError):
            self.ap.logger.warning(f'Invalid {name}: {value!r}, using {default}')
            return default
        if parsed < 1:
            self.ap.logger.warning(f'Invalid {name}: {value!r}, using {default}')
            return default
        return parsed
--- a/src/langbot/pkg/api/http/service/model.py
+++ b/src/langbot/pkg/api/http/service/model.py
@@ -23,6 +23,17 @@ def _parse_provider_api_keys(provider_dict: dict) -> dict:
    return provider_dict
 def _runtime_model_data(model_uuid: str, model_data: dict) -> dict:
    """Return model data for rebuilding runtime models after an update.
    Update payloads intentionally omit uuid before writing to the database.
    Runtime model entities still need the stable uuid so pipeline configs can
    resolve the in-memory model immediately after an edit, without requiring a
    process restart.
    """
    return {**model_data, 'uuid': model_uuid}
 class LLMModelsService:
    ap: app.Application
@@ -173,7 +184,7 @@ class LLMModelsService:
            raise Exception('provider not found')
        runtime_llm_model = await self.ap.model_mgr.load_llm_model_with_provider(
-            persistence_model.LLMModel(**model_data),
+            persistence_model.LLMModel(**_runtime_model_data(model_uuid, model_data)),
            runtime_provider,
        )
        self.ap.model_mgr.llm_models.append(runtime_llm_model)
@@ -334,7 +345,7 @@ class EmbeddingModelsService:
            raise Exception('provider not found')
        runtime_embedding_model = await self.ap.model_mgr.load_embedding_model_with_provider(
-            persistence_model.EmbeddingModel(**model_data),
+            persistence_model.EmbeddingModel(**_runtime_model_data(model_uuid, model_data)),
            runtime_provider,
        )
        self.ap.model_mgr.embedding_models.append(runtime_embedding_model)
@@ -492,7 +503,7 @@ class RerankModelsService:
            raise Exception('provider not found')
        runtime_rerank_model = await self.ap.model_mgr.load_rerank_model_with_provider(
-            persistence_model.RerankModel(**model_data),
+            persistence_model.RerankModel(**_runtime_model_data(model_uuid, model_data)),
            runtime_provider,
        )
        self.ap.model_mgr.rerank_models.append(runtime_rerank_model)
--- a/src/langbot/pkg/api/http/service/monitoring.py
+++ b/src/langbot/pkg/api/http/service/monitoring.py
@@ -18,55 +18,119 @@ class MonitoringService:
    # ========== Cleanup Methods ==========
-    async def cleanup_expired_records(self, retention_days: int) -> dict[str, int]:
+    async def cleanup_expired_records(self, retention_days: int, batch_size: int = 1000) -> dict[str, int]:
        """Delete monitoring records older than the specified retention period.
        Args:
            retention_days: Number of days to retain records.
            batch_size: Maximum rows to delete per table batch.
        Returns:
            A dict mapping table name to the number of deleted rows.
        """
        if retention_days < 1:
            raise ValueError('retention_days must be >= 1')
        if batch_size < 1:
            raise ValueError('batch_size must be >= 1')
        cutoff = datetime.datetime.now(datetime.timezone.utc).replace(tzinfo=None) - datetime.timedelta(
            days=retention_days
        )
-        tables_and_columns: list[tuple[str, type, sqlalchemy.Column]] = [
+        tables_and_columns: list[tuple[str, type, sqlalchemy.Column, sqlalchemy.Column]] = [
            (
                'monitoring_messages',
                persistence_monitoring.MonitoringMessage,
                persistence_monitoring.MonitoringMessage.timestamp,
                persistence_monitoring.MonitoringMessage.id,
            ),
            (
                'monitoring_llm_calls',
                persistence_monitoring.MonitoringLLMCall,
                persistence_monitoring.MonitoringLLMCall.timestamp,
                persistence_monitoring.MonitoringLLMCall.id,
            ),
            (
                'monitoring_embedding_calls',
                persistence_monitoring.MonitoringEmbeddingCall,
                persistence_monitoring.MonitoringEmbeddingCall.timestamp,
                persistence_monitoring.MonitoringEmbeddingCall.id,
            ),
            (
                'monitoring_errors',
                persistence_monitoring.MonitoringError,
                persistence_monitoring.MonitoringError.timestamp,
                persistence_monitoring.MonitoringError.id,
            ),
            (
                'monitoring_sessions',
                persistence_monitoring.MonitoringSession,
                persistence_monitoring.MonitoringSession.last_activity,
                persistence_monitoring.MonitoringSession.session_id,
            ),
            (
                'monitoring_feedback',
                persistence_monitoring.MonitoringFeedback,
                persistence_monitoring.MonitoringFeedback.timestamp,
                persistence_monitoring.MonitoringFeedback.id,
            ),
        ]
        deleted_counts: dict[str, int] = {}
-        for table_name, model_cls, ts_column in tables_and_columns:
+        for table_name, model_cls, ts_column, pk_column in tables_and_columns:
-            result = await self.ap.persistence_mgr.execute_async(sqlalchemy.delete(model_cls).where(ts_column < cutoff))
+            deleted_counts[table_name] = await self._delete_expired_in_batches(
-            deleted_counts[table_name] = result.rowcount
+                model_cls=model_cls,
                ts_column=ts_column,
                pk_column=pk_column,
                cutoff=cutoff,
                batch_size=batch_size,
            )
        if sum(deleted_counts.values()) > 0:
            await self._release_sqlite_space()
        return deleted_counts
    async def _delete_expired_in_batches(
        self,
        model_cls: type,
        ts_column: sqlalchemy.Column,
        pk_column: sqlalchemy.Column,
        cutoff: datetime.datetime,
        batch_size: int,
    ) -> int:
        deleted_total = 0
        while True:
            select_result = await self.ap.persistence_mgr.execute_async(
                sqlalchemy.select(pk_column).where(ts_column < cutoff).limit(batch_size)
            )
            pk_values = list(select_result.scalars().all())
            if not pk_values:
                break
            delete_result = await self.ap.persistence_mgr.execute_async(
                sqlalchemy.delete(model_cls).where(pk_column.in_(pk_values))
            )
            deleted = delete_result.rowcount or 0
            deleted_total += deleted
            if len(pk_values) < batch_size:
                break
        return deleted_total
    async def _release_sqlite_space(self) -> None:
        database_type = self.ap.instance_config.data.get('database', {}).get('use', 'sqlite')
        if database_type != 'sqlite':
            return
        async with self.ap.persistence_mgr.get_db_engine().connect() as conn:
            autocommit_conn = await conn.execution_options(isolation_level='AUTOCOMMIT')
            await autocommit_conn.execute(sqlalchemy.text('PRAGMA wal_checkpoint(TRUNCATE)'))
            await autocommit_conn.execute(sqlalchemy.text('VACUUM'))
    # ========== Recording Methods ==========
    async def record_message(
--- a/src/langbot/pkg/api/http/service/provider.py
+++ b/src/langbot/pkg/api/http/service/provider.py
@@ -17,6 +17,24 @@ class ModelProviderService:
    def __init__(self, ap: app.Application) -> None:
        self.ap = ap
    @staticmethod
    def _normalize_api_keys(api_keys: str | list[str] | tuple[str, ...] | None) -> list[str]:
        if api_keys is None:
            return []
        raw_keys = [api_keys] if isinstance(api_keys, str) else list(api_keys)
        normalized_keys = []
        seen_keys = set()
        for raw_key in raw_keys:
            normalized_key = raw_key.strip() if isinstance(raw_key, str) else ''
            if not normalized_key or normalized_key in seen_keys:
                continue
            normalized_keys.append(normalized_key)
            seen_keys.add(normalized_key)
        return normalized_keys
    async def get_providers(self) -> list[dict]:
        """Get all providers"""
        result = await self.ap.persistence_mgr.execute_async(sqlalchemy.select(persistence_model.ModelProvider))
@@ -59,6 +77,7 @@ class ModelProviderService:
    async def create_provider(self, provider_data: dict) -> str:
        """Create a new provider"""
        provider_data['uuid'] = str(uuid.uuid4())
        provider_data['api_keys'] = self._normalize_api_keys(provider_data.get('api_keys'))
        await self.ap.persistence_mgr.execute_async(
            sqlalchemy.insert(persistence_model.ModelProvider).values(**provider_data)
        )
@@ -72,6 +91,8 @@ class ModelProviderService:
        """Update an existing provider"""
        if 'uuid' in provider_data:
            del provider_data['uuid']
        if 'api_keys' in provider_data:
            provider_data['api_keys'] = self._normalize_api_keys(provider_data.get('api_keys'))
        await self.ap.persistence_mgr.execute_async(
            sqlalchemy.update(persistence_model.ModelProvider)
            .where(persistence_model.ModelProvider.uuid == provider_uuid)
@@ -141,6 +162,8 @@ class ModelProviderService:
    async def find_or_create_provider(self, requester: str, base_url: str, api_keys: list) -> str:
        """Find existing provider or create new one"""
        api_keys = self._normalize_api_keys(api_keys)
        # Try to find existing provider with same config
        result = await self.ap.persistence_mgr.execute_async(
            sqlalchemy.select(persistence_model.ModelProvider).where(
@@ -168,7 +191,7 @@ class ModelProviderService:
                'name': provider_name,
                'requester': requester,
                'base_url': base_url,
-                'api_keys': api_keys or [],
+                'api_keys': api_keys,
            }
        )
@@ -177,7 +200,7 @@ class ModelProviderService:
        await self.ap.persistence_mgr.execute_async(
            sqlalchemy.update(persistence_model.ModelProvider)
            .where(persistence_model.ModelProvider.uuid == '00000000-0000-0000-0000-000000000000')
-            .values(api_keys=[api_key])
+            .values(api_keys=self._normalize_api_keys(api_key))
        )
        await self.ap.model_mgr.reload_provider('00000000-0000-0000-0000-000000000000')
--- a/src/langbot/pkg/core/app.py
+++ b/src/langbot/pkg/core/app.py
@@ -31,6 +31,7 @@ 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 monitoring as monitoring_service
 from ..api.http.service import maintenance as maintenance_service
 from ..discover import engine as discover_engine
 from ..storage import mgr as storagemgr
@@ -155,6 +156,8 @@ class Application:
    monitoring_service: monitoring_service.MonitoringService = None
    maintenance_service: maintenance_service.MaintenanceService = None
    def __init__(self):
        pass
@@ -194,14 +197,30 @@ class Application:
            monitoring_cfg = self.instance_config.data.get('monitoring', {})
            auto_cleanup_cfg = monitoring_cfg.get('auto_cleanup', {})
            if auto_cleanup_cfg.get('enabled', True):
-                retention_days = auto_cleanup_cfg.get('retention_days', 30)
+                retention_days = self._get_positive_int_config(
-                check_interval_hours = auto_cleanup_cfg.get('check_interval_hours', 1)
+                    auto_cleanup_cfg.get('retention_days', 30),
                    default=30,
                    name='monitoring.auto_cleanup.retention_days',
                )
                delete_batch_size = self._get_positive_int_config(
                    auto_cleanup_cfg.get('delete_batch_size', 1000),
                    default=1000,
                    name='monitoring.auto_cleanup.delete_batch_size',
                )
                check_interval_hours = self._get_positive_float_config(
                    auto_cleanup_cfg.get('check_interval_hours', 1),
                    default=1,
                    name='monitoring.auto_cleanup.check_interval_hours',
                )
                async def monitoring_cleanup_loop():
                    check_interval_seconds = check_interval_hours * 3600
                    while True:
                        try:
-                            deleted = await self.monitoring_service.cleanup_expired_records(retention_days)
+                            deleted = await self.monitoring_service.cleanup_expired_records(
                                retention_days,
                                batch_size=delete_batch_size,
                            )
                            total_deleted = sum(deleted.values())
                            if total_deleted > 0:
                                self.logger.info(
@@ -218,6 +237,33 @@ class Application:
                    scopes=[core_entities.LifecycleControlScope.APPLICATION],
                )
            # Start storage/log maintenance task if enabled
            storage_cleanup_cfg = self.instance_config.data.get('storage', {}).get('cleanup', {})
            if storage_cleanup_cfg.get('enabled', True) and self.maintenance_service is not None:
                check_interval_hours = self._get_positive_float_config(
                    storage_cleanup_cfg.get('check_interval_hours', 1),
                    default=1,
                    name='storage.cleanup.check_interval_hours',
                )
                async def storage_cleanup_loop():
                    check_interval_seconds = check_interval_hours * 3600
                    while True:
                        try:
                            deleted = await self.maintenance_service.cleanup_expired_files()
                            total_deleted = sum(deleted.values())
                            if total_deleted > 0:
                                self.logger.info(f'Storage maintenance: deleted expired files: {deleted}')
                        except Exception as e:
                            self.logger.warning(f'Storage maintenance error: {e}')
                        await asyncio.sleep(check_interval_seconds)
                self.task_mgr.create_task(
                    storage_cleanup_loop(),
                    name='storage-maintenance',
                    scopes=[core_entities.LifecycleControlScope.APPLICATION],
                )
            self.task_mgr.create_task(
                never_ending(),
                name='never-ending-task',
@@ -232,6 +278,28 @@ class Application:
            self.logger.error(f'Application runtime fatal exception: {e}')
            self.logger.debug(f'Traceback: {traceback.format_exc()}')
    def _get_positive_int_config(self, value, default: int, name: str) -> int:
        try:
            parsed = int(value)
        except (TypeError, ValueError):
            self.logger.warning(f'Invalid {name}: {value!r}, using {default}')
            return default
        if parsed < 1:
            self.logger.warning(f'Invalid {name}: {value!r}, using {default}')
            return default
        return parsed
    def _get_positive_float_config(self, value, default: float, name: str) -> float:
        try:
            parsed = float(value)
        except (TypeError, ValueError):
            self.logger.warning(f'Invalid {name}: {value!r}, using {default}')
            return default
        if parsed <= 0:
            self.logger.warning(f'Invalid {name}: {value!r}, using {default}')
            return default
        return parsed
    def dispose(self):
        self.plugin_connector.dispose()
--- a/src/langbot/pkg/core/stages/build_app.py
+++ b/src/langbot/pkg/core/stages/build_app.py
@@ -28,6 +28,7 @@ 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 monitoring as monitoring_service
 from ...api.http.service import maintenance as maintenance_service
 from ...discover import engine as discover_engine
 from ...storage import mgr as storagemgr
 from ...utils import logcache
@@ -167,6 +168,9 @@ class BuildAppStage(stage.BootingStage):
        monitoring_service_inst = monitoring_service.MonitoringService(ap)
        ap.monitoring_service = monitoring_service_inst
        maintenance_service_inst = maintenance_service.MaintenanceService(ap)
        ap.maintenance_service = maintenance_service_inst
        async def runtime_disconnect_callback(connector: plugin_connector.PluginRuntimeConnector) -> None:
            await asyncio.sleep(3)
            await plugin_connector_inst.initialize()
--- a/src/langbot/pkg/core/taskmgr.py
+++ b/src/langbot/pkg/core/taskmgr.py
@@ -3,6 +3,7 @@ from __future__ import annotations
 import asyncio
 import typing
 import datetime
 import time
 from . import app
 from . import entities as core_entities
@@ -119,6 +120,7 @@ class TaskWrapper:
        self.label = label if label != '' else name
        self.task.set_name(name)
        self.scopes = scopes
        self.created_at = time.time()
    def assume_exception(self):
        try:
@@ -154,6 +156,7 @@ class TaskWrapper:
            'name': self.name,
            'label': self.label,
            'scopes': [scope.value for scope in self.scopes],
            'created_at': self.created_at,
            'task_context': self.task_context.to_dict(),
            'runtime': {
                'done': self.task.done(),
@@ -193,6 +196,8 @@ class AsyncTaskManager:
    ) -> TaskWrapper:
        wrapper = TaskWrapper(self.ap, coro, task_type, kind, name, label, context, scopes)
        self.tasks.append(wrapper)
        wrapper.task.add_done_callback(lambda _: self._prune_completed_tasks())
        self._prune_completed_tasks()
        return wrapper
    def create_user_task(
@@ -226,6 +231,15 @@ class AsyncTaskManager:
            'id_index': TaskWrapper._id_index,
        }
    def get_stats(self) -> dict:
        completed = sum(1 for t in self.tasks if t.task.done())
        return {
            'total': len(self.tasks),
            'running': len(self.tasks) - completed,
            'completed': completed,
            'id_index': TaskWrapper._id_index,
        }
    def get_task_by_id(self, id: int) -> TaskWrapper | None:
        for t in self.tasks:
            if t.id == id:
@@ -243,3 +257,27 @@ class AsyncTaskManager:
                if not wrapper.task.done():
                    wrapper.task.cancel()
                return
    def _prune_completed_tasks(self):
        completed_limit = (
            self.ap.instance_config.data.get('system', {})
            .get('task_retention', {})
            .get(
                'completed_limit',
                200,
            )
        )
        try:
            completed_limit = int(completed_limit)
        except (TypeError, ValueError):
            completed_limit = 200
        if completed_limit < 1:
            completed_limit = 1
        completed_tasks = [wrapper for wrapper in self.tasks if wrapper.task.done()]
        overflow = len(completed_tasks) - completed_limit
        if overflow <= 0:
            return
        remove_ids = {wrapper.id for wrapper in completed_tasks[:overflow]}
        self.tasks = [wrapper for wrapper in self.tasks if wrapper.id not in remove_ids]
--- a/src/langbot/pkg/pipeline/preproc/preproc.py
+++ b/src/langbot/pkg/pipeline/preproc/preproc.py
@@ -75,6 +75,27 @@ class PreProcessor(stage.PipelineStage):
            query.bot_uuid,
        )
        # Expire externally managed conversation ids after the conversation has
        # been idle for longer than the configured conversation expire time.
        # The idle window is measured from the last preprocess/update time, not
        # from the conversation creation time.
        conversation_expire_time = query.pipeline_config.get('ai', {}).get('runner', {}).get('expire-time', None)
        now = datetime.datetime.now()
        if conversation_expire_time is not None and conversation_expire_time > 0:
            last_update_time = getattr(conversation, 'update_time', None) or getattr(conversation, 'create_time', None)
            if last_update_time is not None:
                conversation_idle_time = now.timestamp() - last_update_time.timestamp()
                if conversation_idle_time > conversation_expire_time:
                    self.ap.logger.info(
                        f'Conversation({query.query_id}) is expired (idle: {conversation_idle_time}s), create new conversation'
                    )
                    conversation.uuid = None
        # Treat every preprocess pass as a conversation activity update. This
        # makes future expiry checks use the latest incoming message/preprocess
        # time instead of the first message/creation time.
        conversation.update_time = now
        # 设置query
        query.session = session
        query.prompt = conversation.prompt.copy()
@@ -160,7 +181,10 @@ class PreProcessor(stage.PipelineStage):
                elif me.url:
                    content_list.append(provider_message.ContentElement.from_file_url(me.url, 'voice'))
            elif isinstance(me, platform_message.File):
-                content_list.append(provider_message.ContentElement.from_file_url(me.url, me.name))
+                if me.base64:
                    content_list.append(provider_message.ContentElement.from_file_base64(me.base64, me.name))
                elif me.url:
                    content_list.append(provider_message.ContentElement.from_file_url(me.url, me.name))
            elif isinstance(me, platform_message.Quote) and quote_msg:
                for msg in me.origin:
                    if isinstance(msg, platform_message.Plain):
@@ -172,7 +196,10 @@ class PreProcessor(stage.PipelineStage):
                            if msg.base64 is not None:
                                content_list.append(provider_message.ContentElement.from_image_base64(msg.base64))
                    elif isinstance(msg, platform_message.File):
-                        content_list.append(provider_message.ContentElement.from_file_url(msg.url, msg.name))
+                        if msg.base64:
                            content_list.append(provider_message.ContentElement.from_file_base64(msg.base64, msg.name))
                        elif msg.url:
                            content_list.append(provider_message.ContentElement.from_file_url(msg.url, msg.name))
                    elif isinstance(msg, platform_message.Voice):
                        if msg.base64:
                            content_list.append(
--- a/src/langbot/pkg/platform/botmgr.py
+++ b/src/langbot/pkg/platform/botmgr.py
@@ -523,7 +523,7 @@ class PlatformManager:
        return None
    async def remove_bot(self, bot_uuid: str):
-        for bot in self.bots:
+        for bot in self.bots[:]:
            if bot.bot_entity.uuid == bot_uuid:
                if bot.enable:
                    await bot.shutdown()
--- a/src/langbot/pkg/platform/sources/dingtalk.yaml
+++ b/src/langbot/pkg/platform/sources/dingtalk.yaml
@@ -19,6 +19,18 @@ spec:
    en: https://link.langbot.app/en/platforms/dingtalk
    ja: https://link.langbot.app/ja/platforms/dingtalk
  config:
    - name: one-click-create
      label:
        en_US: One-Click Create App
        zh_Hans: 一键创建应用
        zh_Hant: 一鍵建立應用
      description:
        en_US: "Scan QR code with DingTalk to automatically create an app and fill in credentials. Note: Robot Code cannot be obtained automatically, you need to copy it from the DingTalk Developer Backend manually."
        zh_Hans: "使用钉钉扫码自动创建应用并填写凭据。注意：机器人代码无法自动获取，需前往钉钉开发者后台手动复制。"
        zh_Hant: "使用釘釘掃碼自動建立應用並填寫憑證。注意：機器人代碼無法自動取得，需前往釘釘開發者後台手動複製。"
      type: qr-code-login
      login_platform: dingtalk
      required: false
    - name: client_id
      label:
        en_US: Client ID
@@ -40,6 +52,10 @@ spec:
        en_US: Robot Code
        zh_Hans: 机器人代码
        zh_Hant: 機器人代碼
      description:
        en_US: "Required for image recognition, file upload and other features. Get it from DingTalk Developer Backend > Robot Configuration."
        zh_Hans: "识图、上传文件等功能必填。请前往钉钉开发者后台 > 机器人配置中获取。"
        zh_Hant: "識圖、上傳檔案等功能必填。請前往釘釘開發者後台 > 機器人設定中取得。"
      type: string
      required: true
      default: ""
--- a/src/langbot/pkg/platform/sources/lark.py
+++ b/src/langbot/pkg/platform/sources/lark.py
@@ -1025,7 +1025,90 @@ class LarkAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter):
        return api_client
    async def send_message(self, target_type: str, target_id: str, message: platform_message.MessageChain):
-        pass
+        text_elements, media_items = await self.message_converter.yiri2target(message, self.api_client)
        # Map standard target_type to Feishu receive_id_type
        if target_type == 'person':
            receive_id_type = 'open_id'
        elif target_type == 'group':
            receive_id_type = 'chat_id'
        else:
            receive_id_type = target_type
        # Send text message if there are text elements
        if text_elements:
            needs_post = any(ele['tag'] == 'at' for paragraph in text_elements for ele in paragraph)
            if needs_post:
                msg_type = 'post'
                final_content = json.dumps(
                    {
                        'zh_Hans': {
                            'title': '',
                            'content': text_elements,
                        },
                    }
                )
            else:
                msg_type = 'text'
                parts = []
                for paragraph in text_elements:
                    para_text = ''.join(ele.get('text', '') for ele in paragraph)
                    if para_text:
                        parts.append(para_text)
                final_content = json.dumps({'text': '\n\n'.join(parts)})
            request: CreateMessageRequest = (
                CreateMessageRequest.builder()
                .receive_id_type(receive_id_type)
                .request_body(
                    CreateMessageRequestBody.builder()
                    .receive_id(target_id)
                    .content(final_content)
                    .msg_type(msg_type)
                    .uuid(str(uuid.uuid4()))
                    .build()
                )
                .build()
            )
            app_access_token = self.get_app_access_token()
            req_opt: RequestOption = (
                RequestOption.builder().app_ticket(self.app_ticket).app_access_token(app_access_token).build()
            )
            response: CreateMessageResponse = self.api_client.im.v1.message.create(request, req_opt)
            if not response.success():
                raise Exception(
                    f'client.im.v1.message.create failed, code: {response.code}, msg: {response.msg}, log_id: {response.get_log_id()}, resp: \n{json.dumps(json.loads(response.raw.content), indent=4, ensure_ascii=False)}'
                )
        # Send media messages separately (image, audio, file, etc.)
        for media in media_items:
            request: CreateMessageRequest = (
                CreateMessageRequest.builder()
                .receive_id_type(receive_id_type)
                .request_body(
                    CreateMessageRequestBody.builder()
                    .receive_id(target_id)
                    .content(json.dumps(media['content']))
                    .msg_type(media['msg_type'])
                    .uuid(str(uuid.uuid4()))
                    .build()
                )
                .build()
            )
            app_access_token = self.get_app_access_token()
            req_opt: RequestOption = (
                RequestOption.builder().app_ticket(self.app_ticket).app_access_token(app_access_token).build()
            )
            response: CreateMessageResponse = self.api_client.im.v1.message.create(request, req_opt)
            if not response.success():
                raise Exception(
                    f'client.im.v1.message.create ({media["msg_type"]}) failed, code: {response.code}, msg: {response.msg}, log_id: {response.get_log_id()}, resp: \n{json.dumps(json.loads(response.raw.content), indent=4, ensure_ascii=False)}'
                )
    async def is_stream_output_supported(self) -> bool:
        is_stream = False
--- a/src/langbot/pkg/platform/sources/lark.yaml
+++ b/src/langbot/pkg/platform/sources/lark.yaml
@@ -23,6 +23,20 @@ spec:
    en: https://link.langbot.app/en/platforms/lark
    ja: https://link.langbot.app/ja/platforms/lark
  config:
    - name: one-click-create
      label:
        en_US: One-Click Create App
        zh_Hans: 一键创建应用
        zh_Hant: 一鍵建立應用
        ja_JP: ワンクリックでアプリ作成
      description:
        en_US: Scan QR code to automatically create a Feishu app and fill in credentials
        zh_Hans: 扫码自动创建飞书应用并填写凭据
        zh_Hant: 掃碼自動建立飛書應用並填寫憑證
        ja_JP: QRコードをスキャンしてFeishuアプリを自動作成し、認証情報を入力
      type: qr-code-login
      login_platform: feishu
      required: false
    - name: app_id
      label:
        en_US: App ID
--- a/src/langbot/pkg/platform/sources/matrix.png
+++ b/src/langbot/pkg/platform/sources/matrix.png
--- a/src/langbot/pkg/platform/sources/matrix.py
+++ b/src/langbot/pkg/platform/sources/matrix.py
@@ -0,0 +1,693 @@
 from __future__ import annotations
 import typing
 import asyncio
 import traceback
 import base64
 import json
 import nio
 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
 import langbot_plugin.api.entities.builtin.platform.entities as platform_entities
 import langbot_plugin.api.definition.abstract.platform.event_logger as abstract_platform_logger
 class MatrixMessageConverter(abstract_platform_adapter.AbstractMessageConverter):
    @staticmethod
    async def yiri2target(message_chain: platform_message.MessageChain, client: nio.AsyncClient) -> list[dict]:
        components = []
        for component in message_chain:
            if isinstance(component, platform_message.Plain):
                components.append({'type': 'text', 'text': component.text})
            elif isinstance(component, platform_message.Image):
                image_bytes = None
                if component.base64:
                    b64_data = component.base64
                    if ';base64,' in b64_data:
                        b64_data = b64_data.split(';base64,', 1)[1]
                    image_bytes = base64.b64decode(b64_data)
                elif component.url:
                    session = httpclient.get_session()
                    async with session.get(component.url) as response:
                        image_bytes = await response.read()
                elif component.path:
                    with open(component.path, 'rb') as f:
                        image_bytes = f.read()
                if image_bytes:
                    resp = await client.upload(image_bytes, content_type='image/png')
                    if isinstance(resp, nio.UploadResponse):
                        components.append({'type': 'image', 'mxc_url': resp.content_uri})
            elif isinstance(component, platform_message.File):
                file_bytes = None
                if component.base64:
                    b64_data = component.base64
                    if ';base64,' in b64_data:
                        b64_data = b64_data.split(';base64,', 1)[1]
                    file_bytes = base64.b64decode(b64_data)
                elif component.url:
                    session = httpclient.get_session()
                    async with session.get(component.url) as response:
                        file_bytes = await response.read()
                elif component.path:
                    with open(component.path, 'rb') as f:
                        file_bytes = f.read()
                if file_bytes:
                    file_name = getattr(component, 'name', None) or 'file'
                    resp = await client.upload(file_bytes, content_type='application/octet-stream', filename=file_name)
                    if isinstance(resp, nio.UploadResponse):
                        components.append(
                            {
                                'type': 'file',
                                'mxc_url': resp.content_uri,
                                'filename': file_name,
                                'size': len(file_bytes),
                            }
                        )
            elif isinstance(component, platform_message.Forward):
                for node in component.node_list:
                    components.extend(await MatrixMessageConverter.yiri2target(node.message_chain, client))
        return components
    @staticmethod
    async def target2yiri(event: nio.RoomMessageText | nio.RoomMessageImage, client: nio.AsyncClient, bot_user_id: str):
        message_components = []
        if isinstance(event, nio.RoomMessageText):
            text = event.body
            if bot_user_id and bot_user_id in text:
                message_components.append(platform_message.At(target=bot_user_id))
                text = text.replace(bot_user_id, '').strip()
            message_components.append(platform_message.Plain(text=text))
        elif isinstance(event, nio.RoomMessageImage):
            mxc_url = event.url
            if mxc_url:
                resp = await client.download(mxc_url)
                if isinstance(resp, nio.DownloadResponse):
                    b64 = base64.b64encode(resp.body).decode('utf-8')
                    content_type = resp.content_type or 'image/png'
                    message_components.append(platform_message.Image(base64=f'data:{content_type};base64,{b64}'))
            if event.body:
                message_components.append(platform_message.Plain(text=event.body))
        return platform_message.MessageChain(message_components)
 class MatrixEventConverter(abstract_platform_adapter.AbstractEventConverter):
    @staticmethod
    async def yiri2target(event: platform_events.MessageEvent):
        return event.source_platform_object
    @staticmethod
    async def target2yiri(
        event: nio.RoomMessageText | nio.RoomMessageImage,
        room: nio.MatrixRoom,
        client: nio.AsyncClient,
        bot_user_id: str,
        bridge_user_ids: list[str] | None = None,
    ):
        lb_message = await MatrixMessageConverter.target2yiri(event, client, bot_user_id)
        # Determine if this is a direct/private chat or a group chat.
        # Exclude bot itself and bridge bots, count remaining real users.
        exclude_ids = {bot_user_id}
        if bridge_user_ids:
            exclude_ids.update(bridge_user_ids)
        real_users = [uid for uid in room.users if uid not in exclude_ids]
        is_direct = len(real_users) <= 1
        if is_direct:
            return platform_events.FriendMessage(
                sender=platform_entities.Friend(
                    id=event.sender,
                    nickname=room.user_name(event.sender) or event.sender,
                    remark='',
                ),
                message_chain=lb_message,
                time=event.server_timestamp / 1000.0,
                source_platform_object={'event': event, 'room': room},
            )
        else:
            return platform_events.GroupMessage(
                sender=platform_entities.GroupMember(
                    id=event.sender,
                    member_name=room.user_name(event.sender) or event.sender,
                    permission=platform_entities.Permission.Member,
                    group=platform_entities.Group(
                        id=room.room_id,
                        name=room.display_name or room.room_id,
                        permission=platform_entities.Permission.Member,
                    ),
                    special_title='',
                ),
                message_chain=lb_message,
                time=event.server_timestamp / 1000.0,
                source_platform_object={'event': event, 'room': room},
            )
 class BridgeState:
    """Per-bridge runtime state."""
    def __init__(self, user_id: str, login_command: str, logout_command: str, success_keyword: str, check_command: str):
        self.user_id = user_id
        self.login_command = login_command
        self.logout_command = logout_command
        self.success_keyword = success_keyword
        self.check_command = check_command or login_command
        self.logged_in = False
        self.dm_room_id: str | None = None
        self.login_task: asyncio.Task | None = None
        self.check_task: asyncio.Task | None = None
        self.check_responded = False
 class MatrixAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter):
    client: typing.Any = None
    message_converter: MatrixMessageConverter = MatrixMessageConverter()
    event_converter: MatrixEventConverter = MatrixEventConverter()
    config: dict
    listeners: typing.Dict[typing.Type[platform_events.Event], typing.Callable] = {}
    _running: bool = False
    _initial_sync_done: bool = False
    _bridges: list = []
    def __init__(self, config: dict, logger: abstract_platform_logger.AbstractEventLogger):
        homeserver_url = config.get('homeserver_url', '')
        access_token = config.get('access_token', '')
        user_id = config.get('user_id', '')
        if not homeserver_url or not access_token or not user_id:
            raise ValueError('Matrix 机器人缺少必要配置项 (homeserver_url, user_id, access_token)')
        client = nio.AsyncClient(homeserver_url, user_id)
        client.access_token = access_token
        client.user_id = user_id
        super().__init__(
            config=config,
            logger=logger,
            bot_account_id=user_id,
            client=client,
            listeners={},
        )
        # Parse bridges config AFTER super().__init__() to avoid Pydantic resetting _bridges
        self._bridges = []
        bridges_raw = config.get('bridges', '')
        if bridges_raw:
            if isinstance(bridges_raw, str):
                try:
                    bridges_list = json.loads(bridges_raw)
                except (json.JSONDecodeError, TypeError) as e:
                    raise ValueError(f'bridges 配置 JSON 解析失败: {e}\n原始值: {bridges_raw}')
            else:
                bridges_list = bridges_raw
            for b in bridges_list:
                if isinstance(b, dict) and b.get('user_id', '').strip():
                    self._bridges.append(
                        BridgeState(
                            user_id=b['user_id'].strip(),
                            login_command=b.get('login_command', '').strip(),
                            logout_command=b.get('logout_command', '').strip(),
                            success_keyword=b.get('success_keyword', 'Successfully logged in').strip(),
                            check_command=b.get('check_command', '').strip(),
                        )
                    )
        # Backward compatibility: old single-bridge config
        if not self._bridges:
            old_user_id = config.get('bridge_user_id', '').strip()
            old_command = config.get('bridge_login_command', '').strip()
            old_keyword = config.get('bridge_login_success_keyword', 'Successfully logged in').strip()
            old_check = config.get('bridge_check_command', '').strip()
            old_logout = config.get('bridge_logout_command', '').strip()
            if old_user_id:
                self._bridges.append(
                    BridgeState(
                        user_id=old_user_id,
                        login_command=old_command,
                        logout_command=old_logout,
                        success_keyword=old_keyword,
                        check_command=old_check,
                    )
                )
    async def send_message(self, target_type: str, target_id: str, message: platform_message.MessageChain):
        components = await self.message_converter.yiri2target(message, self.client)
        for component in components:
            await self._send_component(target_id, component)
    async def reply_message(
        self,
        message_source: platform_events.MessageEvent,
        message: platform_message.MessageChain,
        quote_origin: bool = False,
    ):
        source_obj = message_source.source_platform_object
        room_id = source_obj['room'].room_id
        components = await self.message_converter.yiri2target(message, self.client)
        for component in components:
            if quote_origin:
                original_event = source_obj['event']
                await self._send_component(room_id, component, reply_to=original_event.event_id)
            else:
                await self._send_component(room_id, component)
    async def _send_component(self, room_id: str, component: dict, reply_to: str | None = None):
        content = {}
        if component['type'] == 'text':
            content = {
                'msgtype': 'm.text',
                'body': component['text'],
            }
        elif component['type'] == 'image':
            content = {
                'msgtype': 'm.image',
                'body': 'image.png',
                'url': component['mxc_url'],
            }
        elif component['type'] == 'file':
            content = {
                'msgtype': 'm.file',
                'body': component.get('filename', 'file'),
                'url': component['mxc_url'],
                'info': {'size': component.get('size', 0)},
            }
        if reply_to and content:
            content['m.relates_to'] = {
                'm.in_reply_to': {'event_id': reply_to},
            }
        if content:
            await self.client.room_send(
                room_id=room_id,
                message_type='m.room.message',
                content=content,
            )
    def register_listener(
        self,
        event_type: typing.Type[platform_events.Event],
        callback: typing.Callable[
            [platform_events.Event, abstract_platform_adapter.AbstractMessagePlatformAdapter], None
        ],
    ):
        self.listeners[event_type] = callback
    async def run_async(self):
        self._running = True
        await self.logger.info('Matrix adapter starting...')
        # Debug: log bridge parsing result
        bridges_raw = self.config.get('bridges', '')
        await self.logger.debug(f'bridges config raw: type={type(bridges_raw).__name__}, repr={repr(bridges_raw)}')
        await self.logger.debug(
            f'parsed _bridges count: {len(self._bridges)}, ids: {[b.user_id for b in self._bridges]}'
        )
        # Collect all bridge bot user IDs for filtering
        _bridge_user_ids = [b.user_id for b in self._bridges]
        _bridge_user_id_set = set(_bridge_user_ids)
        # Auto-join invited rooms
        async def on_invite(room: nio.MatrixRoom, event: nio.InviteMemberEvent):
            if event.membership == 'invite' and event.state_key == self.client.user_id:
                await self.client.join(room.room_id)
                await self.logger.debug(f'Auto-joined room: {room.display_name or room.room_id}')
        self.client.add_event_callback(on_invite, nio.InviteMemberEvent)
        # Handle text messages
        async def on_message(room: nio.MatrixRoom, event: nio.RoomMessageText):
            if not self._initial_sync_done:
                return
            if event.sender == self.client.user_id:
                return
            # Admin commands (from any non-bridge user)
            if event.sender not in _bridge_user_id_set:
                body = (event.body or '').strip()
                if body == '!relogin':
                    await self._handle_relogin_command(room.room_id)
                    return
                if body == '!status':
                    await self._handle_status_command(room.room_id)
                    return
            if event.sender in _bridge_user_id_set:
                return
            try:
                lb_event = await self.event_converter.target2yiri(
                    event, room, self.client, self.bot_account_id, _bridge_user_ids
                )
                if type(lb_event) in self.listeners:
                    result = self.listeners[type(lb_event)](lb_event, self)
                    if asyncio.iscoroutine(result):
                        await result
            except Exception:
                await self.logger.error(f'Error handling Matrix message: {traceback.format_exc()}')
        self.client.add_event_callback(on_message, nio.RoomMessageText)
        # Handle image messages
        async def on_image(room: nio.MatrixRoom, event: nio.RoomMessageImage):
            if not self._initial_sync_done:
                return
            if event.sender == self.client.user_id:
                return
            if event.sender in _bridge_user_id_set:
                return
            try:
                lb_event = await self.event_converter.target2yiri(
                    event, room, self.client, self.bot_account_id, _bridge_user_ids
                )
                if type(lb_event) in self.listeners:
                    result = self.listeners[type(lb_event)](lb_event, self)
                    if asyncio.iscoroutine(result):
                        await result
            except Exception:
                await self.logger.error(f'Error handling Matrix image: {traceback.format_exc()}')
        self.client.add_event_callback(on_image, nio.RoomMessageImage)
        # Set up bridge-specific callbacks for each bridge
        _disconnect_keywords = ['disconnected', 'logged out', 'connection lost', 'session expired', 'token expired']
        for bridge in self._bridges:
            # Login success detection (notice)
            async def on_bridge_notice(room: nio.MatrixRoom, event: nio.RoomMessageNotice, _b=bridge):
                if not self._initial_sync_done:
                    return
                if event.sender != _b.user_id:
                    return
                _b.check_responded = True
                if _b.success_keyword in (event.body or ''):
                    _b.logged_in = True
                    await self.logger.info(f'[{_b.user_id}] Bridge login succeeded.')
                # Disconnect detection
                body_lower = (event.body or '').lower()
                for kw in _disconnect_keywords:
                    if kw in body_lower and _b.logged_in:
                        _b.logged_in = False
                        await self.logger.info(f'[{_b.user_id}] Bridge 账号掉线 (检测到: "{kw}"), 将自动重新登录...')
                        self._restart_bridge_login(_b)
                        break
            self.client.add_event_callback(on_bridge_notice, nio.RoomMessageNotice)
            # Login success + disconnect detection (text)
            async def on_bridge_text(room: nio.MatrixRoom, event: nio.RoomMessageText, _b=bridge):
                if not self._initial_sync_done:
                    return
                if event.sender != _b.user_id:
                    return
                _b.check_responded = True
                if _b.success_keyword in (event.body or ''):
                    _b.logged_in = True
                    await self.logger.info(f'[{_b.user_id}] Bridge login succeeded.')
                body_lower = (event.body or '').lower()
                for kw in _disconnect_keywords:
                    if kw in body_lower and _b.logged_in:
                        _b.logged_in = False
                        await self.logger.info(f'[{_b.user_id}] Bridge 账号掉线 (检测到: "{kw}"), 将自动重新登录...')
                        self._restart_bridge_login(_b)
                        break
            self.client.add_event_callback(on_bridge_text, nio.RoomMessageText)
            # QR code image forwarding
            async def on_bridge_image(room: nio.MatrixRoom, event: nio.RoomMessageImage, _b=bridge):
                if not self._initial_sync_done:
                    return
                if event.sender != _b.user_id:
                    return
                mxc_url = event.url
                if not mxc_url:
                    return
                try:
                    resp = await self.client.download(mxc_url)
                    if isinstance(resp, nio.DownloadResponse):
                        b64 = base64.b64encode(resp.body).decode('utf-8')
                        content_type = resp.content_type or 'image/png'
                        await self.logger.info(
                            f'[{_b.user_id}] Bridge 发送了二维码，请扫码登录:',
                            images=[platform_message.Image(base64=f'data:{content_type};base64,{b64}')],
                        )
                except Exception:
                    await self.logger.error(
                        f'[{_b.user_id}] Failed to download bridge QR image: {traceback.format_exc()}'
                    )
            self.client.add_event_callback(on_bridge_image, nio.RoomMessageImage)
        await self.logger.debug('Matrix adapter running, starting sync...')
        # Initial sync to skip old messages
        resp = await self.client.sync(timeout=10000)
        if isinstance(resp, nio.SyncResponse):
            await self.logger.debug(f'Matrix initial sync done, next_batch: {resp.next_batch}')
        self._initial_sync_done = True
        # Display account info
        display_name = self.client.user_id
        try:
            profile_resp = await self.client.get_displayname(self.client.user_id)
            if isinstance(profile_resp, nio.ProfileGetDisplayNameResponse) and profile_resp.displayname:
                display_name = profile_resp.displayname
        except Exception:
            pass
        joined_rooms = len(self.client.rooms)
        homeserver = self.config.get('homeserver_url', '')
        bridge_info = ''
        if self._bridges:
            bridge_names = ', '.join(b.user_id for b in self._bridges)
            bridge_info = f' | 桥接: [{bridge_names}]'
        await self.logger.info(
            f'Matrix 账号: {display_name} ({self.client.user_id}) | '
            f'服务器: {homeserver} | 已加入 {joined_rooms} 个房间{bridge_info}'
        )
        # Start bridge login and status check tasks for each bridge
        for bridge in self._bridges:
            if bridge.login_command:
                await self.logger.info(
                    f'[{bridge.user_id}] Bridge login enabled (命令: "{bridge.login_command}", '
                    f'关键词: "{bridge.success_keyword}")'
                )
                bridge.login_task = asyncio.create_task(self._periodic_bridge_login(bridge))
                bridge.check_task = asyncio.create_task(self._periodic_bridge_check(bridge))
            else:
                await self.logger.debug(f'[{bridge.user_id}] Bridge login not configured (no login_command)')
        # Main sync loop
        while self._running:
            try:
                await self.client.sync(timeout=30000)
            except Exception:
                await self.logger.error(f'Matrix sync error: {traceback.format_exc()}')
                await asyncio.sleep(5)
    async def _periodic_bridge_login(self, bridge: BridgeState):
        """Periodically send login command to a bridge bot until login succeeds."""
        try:
            await self.logger.info(f'[{bridge.user_id}] Bridge login task started, looking for DM room...')
            dm_room_id = None
            for room_id, room in self.client.rooms.items():
                if room.member_count == 2 and bridge.user_id in [m for m in room.users]:
                    dm_room_id = room_id
                    break
            if not dm_room_id:
                resp = await self.client.room_create(
                    is_direct=True,
                    invite=[bridge.user_id],
                )
                if isinstance(resp, nio.RoomCreateResponse):
                    dm_room_id = resp.room_id
                    await self.logger.debug(f'[{bridge.user_id}] Created DM room: {dm_room_id}')
                else:
                    await self.logger.error(f'[{bridge.user_id}] Failed to create DM room: {resp}')
                    return
            bridge.dm_room_id = dm_room_id
            # Force logout first on every adapter start
            logout_cmd = bridge.logout_command or bridge.login_command.replace('login', 'logout')
            await self.logger.info(f'[{bridge.user_id}] 强制登出: "{logout_cmd}"')
            await self.client.room_send(
                room_id=dm_room_id,
                message_type='m.room.message',
                content={'msgtype': 'm.text', 'body': logout_cmd},
            )
            await asyncio.sleep(3)
            while self._running and not bridge.logged_in:
                await self.logger.debug(f'[{bridge.user_id}] Sending "{bridge.login_command}" in room {dm_room_id}')
                await self.client.room_send(
                    room_id=dm_room_id,
                    message_type='m.room.message',
                    content={'msgtype': 'm.text', 'body': bridge.login_command},
                )
                for _ in range(60):
                    if not self._running or bridge.logged_in:
                        break
                    await asyncio.sleep(1)
            if bridge.logged_in:
                await self.logger.debug(f'[{bridge.user_id}] Bridge login confirmed, periodic login stopped.')
        except asyncio.CancelledError:
            pass
        except Exception:
            await self.logger.error(f'[{bridge.user_id}] Bridge periodic login error: {traceback.format_exc()}')
    def _restart_bridge_login(self, bridge: BridgeState):
        """Cancel existing login task and start a new one."""
        if bridge.login_task and not bridge.login_task.done():
            bridge.login_task.cancel()
        bridge.login_task = asyncio.create_task(self._periodic_bridge_login(bridge))
    async def _periodic_bridge_check(self, bridge: BridgeState):
        """Periodically check a bridge's login status."""
        try:
            while self._running and not bridge.logged_in:
                await asyncio.sleep(5)
            check_interval = 300  # 5 minutes
            response_timeout = 30
            await self.logger.debug(f'[{bridge.user_id}] Bridge status check started (interval: {check_interval}s)')
            while self._running:
                for _ in range(check_interval):
                    if not self._running:
                        return
                    await asyncio.sleep(1)
                if not bridge.logged_in or not bridge.dm_room_id:
                    continue
                try:
                    bridge.check_responded = False
                    await self.client.room_send(
                        room_id=bridge.dm_room_id,
                        message_type='m.room.message',
                        content={'msgtype': 'm.text', 'body': bridge.check_command},
                    )
                    await self.logger.debug(f'[{bridge.user_id}] Bridge status check: sent "{bridge.check_command}"')
                    for _ in range(response_timeout):
                        if bridge.check_responded or not self._running:
                            break
                        await asyncio.sleep(1)
                    if bridge.check_responded:
                        await self.logger.debug(f'[{bridge.user_id}] Bridge status check: OK')
                    else:
                        await self.logger.info(
                            f'[{bridge.user_id}] Bridge status check: 无响应, 可能已掉线, 尝试重新登录...'
                        )
                        bridge.logged_in = False
                        self._restart_bridge_login(bridge)
                except Exception:
                    await self.logger.error(f'[{bridge.user_id}] Bridge status check error: {traceback.format_exc()}')
        except asyncio.CancelledError:
            pass
        except Exception:
            await self.logger.error(f'[{bridge.user_id}] Bridge status check fatal error: {traceback.format_exc()}')
    async def _handle_relogin_command(self, room_id: str):
        """Handle !relogin command: logout then re-login all bridges."""
        if not self._bridges:
            await self.client.room_send(
                room_id=room_id,
                message_type='m.room.message',
                content={'msgtype': 'm.text', 'body': '没有配置任何桥。'},
            )
            return
        lines = ['开始重新登录所有桥...']
        for bridge in self._bridges:
            if not bridge.login_command or not bridge.dm_room_id:
                lines.append(f'[{bridge.user_id}] 跳过（未配置登录命令或无DM房间）')
                continue
                # Use configured logout command, fallback to deriving from login command
                logout_cmd = bridge.logout_command or bridge.login_command.replace('login', 'logout')
            lines.append(f'[{bridge.user_id}] 发送 "{logout_cmd}"...')
            # Cancel existing tasks
            if bridge.login_task and not bridge.login_task.done():
                bridge.login_task.cancel()
            if bridge.check_task and not bridge.check_task.done():
                bridge.check_task.cancel()
            # Send logout
            try:
                await self.client.room_send(
                    room_id=bridge.dm_room_id,
                    message_type='m.room.message',
                    content={'msgtype': 'm.text', 'body': logout_cmd},
                )
            except Exception as e:
                lines.append(f'[{bridge.user_id}] logout 发送失败: {e}')
            await asyncio.sleep(2)
            # Reset state and restart login
            bridge.logged_in = False
            self._restart_bridge_login(bridge)
            lines.append(f'[{bridge.user_id}] 已触发重新登录')
        await self.client.room_send(
            room_id=room_id,
            message_type='m.room.message',
            content={'msgtype': 'm.text', 'body': '\n'.join(lines)},
        )
    async def _handle_status_command(self, room_id: str):
        """Handle !status command: show bridge states."""
        if not self._bridges:
            await self.client.room_send(
                room_id=room_id,
                message_type='m.room.message',
                content={'msgtype': 'm.text', 'body': '没有配置任何桥。'},
            )
            return
        lines = ['桥状态:']
        for bridge in self._bridges:
            status = '已登录 ✓' if bridge.logged_in else '未登录 ✗'
            dm = bridge.dm_room_id or '无'
            lines.append(f'• {bridge.user_id}: {status} (DM: {dm})')
        await self.client.room_send(
            room_id=room_id,
            message_type='m.room.message',
            content={'msgtype': 'm.text', 'body': '\n'.join(lines)},
        )
    async def kill(self) -> bool:
        self._running = False
        for bridge in self._bridges:
            if bridge.login_task and not bridge.login_task.done():
                bridge.login_task.cancel()
            if bridge.check_task and not bridge.check_task.done():
                bridge.check_task.cancel()
        if self.client:
            await self.client.close()
        await self.logger.debug('Matrix adapter stopped')
        return True
    async def unregister_listener(
        self,
        event_type: typing.Type[platform_events.Event],
        callback: typing.Callable[
            [platform_events.Event, abstract_platform_adapter.AbstractMessagePlatformAdapter], None
        ],
    ):
        if event_type in self.listeners:
            del self.listeners[event_type]
--- a/src/langbot/pkg/platform/sources/matrix.yaml
+++ b/src/langbot/pkg/platform/sources/matrix.yaml
@@ -0,0 +1,123 @@
 apiVersion: v1
 kind: MessagePlatformAdapter
 metadata:
  name: matrix
  label:
    en_US: Matrix
    zh_Hans: Matrix
    zh_Hant: Matrix
    ja_JP: Matrix
    th_TH: Matrix
    vi_VN: Matrix
    es_ES: Matrix
  description:
    en_US: Matrix protocol adapter, supports self-hosted Synapse servers and any Matrix-compatible homeserver
    zh_Hans: Matrix 协议适配器，支持自建 Synapse 服务器及任何 Matrix 兼容的 Homeserver
    zh_Hant: Matrix 協議適配器，支持自建 Synapse 伺服器及任何 Matrix 相容的 Homeserver
    ja_JP: Matrix プロトコルアダプター、セルフホストの Synapse サーバーおよび Matrix 互換のホームサーバーをサポート
    th_TH: อะแดปเตอร์โปรโตคอล Matrix รองรับเซิร์ฟเวอร์ Synapse ที่โฮสต์เองและ Homeserver ที่เข้ากันได้กับ Matrix
    vi_VN: Bộ điều hợp giao thức Matrix, hỗ trợ máy chủ Synapse tự lưu trữ và bất kỳ Homeserver tương thích Matrix nào
    es_ES: Adaptador del protocolo Matrix, compatible con servidores Synapse autoalojados y cualquier Homeserver compatible con Matrix
  icon: matrix.png
 spec:
  categories:
    - global
    - protocol
  config:
    - name: homeserver_url
      label:
        en_US: Homeserver URL
        zh_Hans: Homeserver 地址
        zh_Hant: Homeserver 地址
        ja_JP: Homeserver URL
        th_TH: URL ของ Homeserver
        vi_VN: URL Homeserver
        es_ES: URL del Homeserver
      description:
        en_US: "The URL of the Matrix homeserver, e.g. http://localhost:8008"
        zh_Hans: "Matrix Homeserver 的地址，例如 http://localhost:8008"
      type: string
      required: true
      default: "http://localhost:8008"
    - name: user_id
      label:
        en_US: Bot User ID
        zh_Hans: 机器人用户 ID
        zh_Hant: 機器人用戶 ID
        ja_JP: ボットユーザー ID
        th_TH: ID ผู้ใช้บอท
        vi_VN: ID người dùng bot
        es_ES: ID de usuario del bot
      description:
        en_US: "The full Matrix user ID, e.g. @bot:localhost"
        zh_Hans: "完整的 Matrix 用户 ID，例如 @bot:localhost"
      type: string
      required: true
      default: "@langbot:localhost"
    - name: access_token
      label:
        en_US: Access Token
        zh_Hans: 访问令牌
        zh_Hant: 訪問令牌
        ja_JP: アクセストークン
        th_TH: โทเค็นการเข้าถึง
        vi_VN: Mã truy cập
        es_ES: Token de acceso
      description:
        en_US: "Access token obtained by logging in via the Matrix client API"
        zh_Hans: "通过 Matrix Client API 登录获取的访问令牌"
      type: string
      required: true
      default: ""
    - name: bridge_user_id
      label:
        en_US: Bridge Bot User ID (single bridge, legacy)
        zh_Hans: 桥机器人用户 ID（单桥兼容）
      description:
        en_US: "Single bridge bot user ID (legacy). Prefer 'bridges' for multi-bridge. e.g. @discordbot:localhost"
        zh_Hans: "单桥机器人用户 ID（旧格式兼容）。推荐使用 bridges 配置多桥。例如 @discordbot:localhost"
      type: string
      required: false
      default: ""
    - name: bridge_login_command
      label:
        en_US: Bridge Login Command (single bridge, legacy)
        zh_Hans: 桥登录命令（单桥兼容）
      description:
        en_US: "Login command for single bridge (legacy). e.g. !discord login"
        zh_Hans: "单桥登录命令（旧格式兼容）。例如 !discord login"
      type: string
      required: false
      default: ""
    - name: bridge_login_success_keyword
      label:
        en_US: Bridge Login Success Keyword (single bridge, legacy)
        zh_Hans: 桥登录成功关键词（单桥兼容）
      description:
        en_US: "Success keyword for single bridge (legacy). e.g. Successfully logged in"
        zh_Hans: "单桥登录成功关键词（旧格式兼容）。例如 Successfully logged in"
      type: string
      required: false
      default: "Successfully logged in"
    - name: bridges
      label:
        en_US: Bridges Config (Multi-bridge)
        zh_Hans: 桥配置（多桥）
      description:
        en_US: >
          JSON array of bridge configs. Each bridge: {"user_id": "@bot:host", "login_command": "!xx login",
          "success_keyword": "logged in", "check_command": "!xx ping"}.
          Example: [{"user_id":"@discordbot:localhost","login_command":"!discord login","success_keyword":"logged in"},
          {"user_id":"@telegrambot:localhost","login_command":"!tg login","success_keyword":"logged in"}]
        zh_Hans: >
          JSON 数组格式的多桥配置。每个桥: {"user_id": "@bot:host", "login_command": "!xx login",
          "success_keyword": "logged in", "check_command": "!xx ping"}。
          示例: [{"user_id":"@discordbot:localhost","login_command":"!discord login","success_keyword":"logged in"},
          {"user_id":"@telegrambot:localhost","login_command":"!tg login","success_keyword":"logged in"}]
      type: string
      required: false
      default: ""
 execution:
  python:
    path: ./matrix.py
    attr: MatrixAdapter
--- a/src/langbot/pkg/platform/sources/openclaw_weixin.yaml
+++ b/src/langbot/pkg/platform/sources/openclaw_weixin.yaml
@@ -32,6 +32,20 @@ spec:
      type: string
      required: true
      default: "https://ilinkai.weixin.qq.com"
    - name: qr-login
      label:
        en_US: Scan QR Login
        zh_Hans: 扫码登录
        zh_Hant: 掃碼登入
        ja_JP: QRコードでログイン
      description:
        en_US: Scan QR code with WeChat to authorize and automatically fill in the token
        zh_Hans: 使用微信扫码授权，自动填写令牌
        zh_Hant: 使用微信掃碼授權，自動填寫令牌
        ja_JP: WeChatでQRコードをスキャンし、トークンを自動入力
      type: qr-code-login
      login_platform: weixin
      required: false
    - name: token
      label:
        en_US: Token
--- a/src/langbot/pkg/platform/sources/qqofficial.py
+++ b/src/langbot/pkg/platform/sources/qqofficial.py
@@ -1,9 +1,11 @@
 from __future__ import annotations
 import typing
 import re
 import asyncio
 import traceback
 import datetime
 import time
 import langbot_plugin.api.definition.abstract.platform.adapter as abstract_platform_adapter
 import langbot_plugin.api.entities.builtin.platform.message as platform_message
@@ -15,11 +17,25 @@ from ...utils import image
 from ..logger import EventLogger
 def _is_base64_data(value: str) -> bool:
    """Check if a string contains base64-encoded data rather than a URL."""
    if not value:
        return False
    # data: URI scheme (e.g. data:image/png;base64,xxx)
    if value.startswith('data:'):
        return True
    # Only treat as base64 if it doesn't look like a URL/path and has valid base64 chars
    if value.startswith(('http://', 'https://', '/', './', '../')):
        return False
    # Check if it looks like base64 (only valid chars, reasonable length)
    return bool(re.fullmatch(r'[A-Za-z0-9+/=\s]{20,}', value))
 class QQOfficialMessageConverter(abstract_platform_adapter.AbstractMessageConverter):
    @staticmethod
    async def yiri2target(message_chain: platform_message.MessageChain):
        """将 LangBot 消息链转换为 QQ Official 消息格式列表。"""
        content_list = []
        # 只实现了发文字
        for msg in message_chain:
            if type(msg) is platform_message.Plain:
                content_list.append(
@@ -28,6 +44,49 @@ class QQOfficialMessageConverter(abstract_platform_adapter.AbstractMessageConver
                        'content': msg.text,
                    }
                )
            elif type(msg) is platform_message.Image:
                url = msg.url if hasattr(msg, 'url') and msg.url else None
                b64 = msg.base64 if hasattr(msg, 'base64') and msg.base64 else None
                # Some plugins (e.g. MimoTTS) store base64 data in the url field
                if url and not b64 and _is_base64_data(url):
                    b64 = url
                    url = None
                content_list.append(
                    {
                        'type': 'image',
                        'url': url,
                        'base64': b64,
                    }
                )
            elif type(msg) is platform_message.Voice:
                url = msg.url if hasattr(msg, 'url') and msg.url else None
                b64 = msg.base64 if hasattr(msg, 'base64') and msg.base64 else None
                # Some plugins (e.g. MimoTTS) store base64 data in the url field
                if url and not b64 and _is_base64_data(url):
                    b64 = url
                    url = None
                content_list.append(
                    {
                        'type': 'voice',
                        'url': url,
                        'base64': b64,
                    }
                )
            elif type(msg) is platform_message.File:
                url = msg.url if hasattr(msg, 'url') and msg.url else None
                b64 = msg.base64 if hasattr(msg, 'base64') and msg.base64 else None
                # Some plugins store base64 data in the url field
                if url and not b64 and _is_base64_data(url):
                    b64 = url
                    url = None
                content_list.append(
                    {
                        'type': 'file',
                        'url': url,
                        'base64': b64,
                        'name': msg.name if hasattr(msg, 'name') else 'file',
                    }
                )
        return content_list
@@ -129,12 +188,19 @@ class QQOfficialAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter
    config: dict
    bot_account_id: str
    bot_uuid: str = None
    enable_webhook: bool = False
    message_converter: QQOfficialMessageConverter = QQOfficialMessageConverter()
    event_converter: QQOfficialEventConverter = QQOfficialEventConverter()
    def __init__(self, config: dict, logger: EventLogger):
        enable_webhook = config.get('enable-webhook', False)
        bot = QQOfficialClient(
-            app_id=config['appid'], secret=config['secret'], token=config['token'], logger=logger, unified_mode=True
+            app_id=config['appid'],
            secret=config['secret'],
            token=config['token'],
            logger=logger,
            unified_mode=enable_webhook,
        )
        super().__init__(
@@ -144,6 +210,13 @@ class QQOfficialAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter
            bot_account_id=config['appid'],
        )
        self.enable_webhook = enable_webhook
        self._ws_task: asyncio.Task = None
        self._stream_ctx: dict = {}
        self._stream_ctx_ts: dict[str, float] = {}
        self._fallback_text: dict[str, str] = {}
        self._fallback_text_ts: dict[str, float] = {}
    async def reply_message(
        self,
        message_source: platform_events.MessageEvent,
@@ -156,28 +229,18 @@ class QQOfficialAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter
        content_list = await QQOfficialMessageConverter.yiri2target(message)
-        # 私聊消息
+        # 确定 target_type 和 target_id
        target_type = None
        target_id = None
        if qq_official_event.t == 'C2C_MESSAGE_CREATE':
-            for content in content_list:
+            target_type = 'c2c'
-                if content['type'] == 'text':
+            target_id = qq_official_event.user_openid
-                    await self.bot.send_private_text_msg(
+        elif qq_official_event.t == 'GROUP_AT_MESSAGE_CREATE':
-                        qq_official_event.user_openid,
+            target_type = 'group'
-                        content['content'],
+            target_id = qq_official_event.group_openid
-                        qq_official_event.d_id,
+        elif qq_official_event.t == 'AT_MESSAGE_CREATE':
-                    )
+            # 频道群聊使用频道 API，暂不支持富媒体
        # 群聊消息
        if qq_official_event.t == 'GROUP_AT_MESSAGE_CREATE':
            for content in content_list:
                if content['type'] == 'text':
                    await self.bot.send_group_text_msg(
                        qq_official_event.group_openid,
                        content['content'],
                        qq_official_event.d_id,
                    )
        # 频道群聊
        if qq_official_event.t == 'AT_MESSAGE_CREATE':
            for content in content_list:
                if content['type'] == 'text':
                    await self.bot.send_channle_group_text_msg(
@@ -185,9 +248,9 @@ class QQOfficialAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter
                        content['content'],
                        qq_official_event.d_id,
                    )
-
+            return
-        # 频道私聊
+        elif qq_official_event.t == 'DIRECT_MESSAGE_CREATE':
-        if qq_official_event.t == 'DIRECT_MESSAGE_CREATE':
+            # 频道私聊使用频道 API，暂不支持富媒体
            for content in content_list:
                if content['type'] == 'text':
                    await self.bot.send_channle_private_text_msg(
@@ -195,6 +258,63 @@ class QQOfficialAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter
                        content['content'],
                        qq_official_event.d_id,
                    )
            return
        # C2C 和群聊：支持文字 + 富媒体
        for content in content_list:
            content_type = content.get('type', 'text')
            if content_type == 'text':
                if target_type == 'c2c':
                    await self.bot.send_private_text_msg(
                        target_id,
                        content['content'],
                        qq_official_event.d_id,
                    )
                elif target_type == 'group':
                    await self.bot.send_group_text_msg(
                        target_id,
                        content['content'],
                        qq_official_event.d_id,
                    )
            elif content_type == 'image':
                file_url = content.get('url')
                file_data = content.get('base64')
                if file_url or file_data:
                    await self.bot.send_image_msg(
                        target_type,
                        target_id,
                        file_url=file_url,
                        file_data=file_data,
                        msg_id=qq_official_event.d_id,
                    )
            elif content_type == 'voice':
                file_url = content.get('url')
                file_data = content.get('base64')
                if file_url or file_data:
                    await self.bot.send_voice_msg(
                        target_type,
                        target_id,
                        file_url=file_url,
                        file_data=file_data,
                        msg_id=qq_official_event.d_id,
                    )
            elif content_type == 'file':
                file_url = content.get('url')
                file_data = content.get('base64')
                file_name = content.get('name', 'file')
                if file_url or file_data:
                    await self.bot.send_file_msg(
                        target_type,
                        target_id,
                        file_url=file_url,
                        file_data=file_data,
                        file_name=file_name,
                        msg_id=qq_official_event.d_id,
                    )
    async def send_message(self, target_type: str, target_id: str, message: platform_message.MessageChain):
        pass
@@ -238,17 +358,196 @@ class QQOfficialAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter
        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
+        if not self.enable_webhook:
-        # 保持运行但不启动独立端口
+            await self._run_websocket()
        else:
            # 统一 webhook 模式下，不启动独立的 Quart 应用
            async def keep_alive():
                while True:
                    await asyncio.sleep(1)
-        async def keep_alive():
+            await keep_alive()
            while True:
                await asyncio.sleep(1)
-        await keep_alive()
+    async def _run_websocket(self):
        """以 WebSocket 模式运行网关连接"""
        await self.logger.info('QQ Official adapter starting in WebSocket mode')
        async def on_ready():
            await self.logger.info('QQ Official WebSocket connected and ready')
        async def on_event(event_type: str, event_data: dict):
            # 只处理消息事件，忽略 READY/RESUMED 等系统事件
            message_event_types = {
                'C2C_MESSAGE_CREATE',
                'DIRECT_MESSAGE_CREATE',
                'GROUP_AT_MESSAGE_CREATE',
                'AT_MESSAGE_CREATE',
            }
            if event_type not in message_event_types:
                return
            if not isinstance(event_data, dict):
                await self.logger.warning(f'Event data is not dict, skipping: {event_type} -> {type(event_data)}')
                return
            await self.logger.info(f'Processing message event: {event_type}')
            # 构造与 webhook 模式相同的 payload 结构
            payload = {'t': event_type, 'd': event_data}
            message_data = await self.bot.get_message(payload)
            if message_data:
                event = QQOfficialEvent.from_payload(message_data)
                await self.bot._handle_message(event)
        async def on_error(error: Exception):
            await self.logger.error(f'WebSocket error: {error}')
            await self.logger.error(f'QQ Official WebSocket error: {error}')
        self._ws_task = asyncio.create_task(self.bot.connect_gateway_loop(on_event, on_ready, on_error))
        try:
            await self._ws_task
        except asyncio.CancelledError:
            pass
    async def kill(self) -> bool:
-        return False
+        if self._ws_task:
            self._ws_task.cancel()
            try:
                await self._ws_task
            except asyncio.CancelledError:
                pass
            self._ws_task = None
        return True
    # --------------- 流式输出 ---------------
    _STREAM_CTX_TTL = 300  # seconds
    async def _cleanup_stale_streams(self):
        """Remove stream contexts that have not been updated for more than _STREAM_CTX_TTL seconds."""
        now = time.time()
        stale_ids = [mid for mid, ts in self._stream_ctx_ts.items() if now - ts > self._STREAM_CTX_TTL]
        for mid in stale_ids:
            self._stream_ctx.pop(mid, None)
            self._stream_ctx_ts.pop(mid, None)
        stale_fb = [mid for mid, ts in self._fallback_text_ts.items() if now - ts > self._STREAM_CTX_TTL]
        for mid in stale_fb:
            self._fallback_text.pop(mid, None)
            self._fallback_text_ts.pop(mid, None)
        if stale_ids or stale_fb:
            await self.logger.debug(f'Cleaned up {len(stale_ids)} stream contexts, {len(stale_fb)} fallback texts')
    async def is_stream_output_supported(self) -> bool:
        return self.config.get('enable-stream-reply', False)
    async def create_message_card(self, message_id: str, event: platform_events.MessageEvent) -> bool:
        source = event.source_platform_object
        # Streaming API only supports C2C private chat
        if source.t != 'C2C_MESSAGE_CREATE':
            return False
        ctx = {
            'user_openid': source.user_openid,
            'msg_id': source.d_id,
            'stream_msg_id': None,
            'msg_seq': 1,
            'index': 0,
            'last_update_ts': 0,
            'accumulated_text': '',
            'sent_length': 0,
            'session_started': False,
        }
        self._stream_ctx[message_id] = ctx
        self._stream_ctx_ts[message_id] = time.time()
        return True
    async def reply_message_chunk(
        self,
        message_source: platform_events.MessageEvent,
        bot_message: dict,
        message: platform_message.MessageChain,
        quote_origin: bool = False,
        is_final: bool = False,
    ):
        # Periodically clean up stale stream contexts
        await self._cleanup_stale_streams()
        # 提取纯文本内容（当前 chunk 的文本）
        text_parts = []
        for msg in message:
            if type(msg) is platform_message.Plain:
                text_parts.append(msg.text)
        chunk_text = '\n\n'.join(text_parts)
        message_id = (
            bot_message.get('resp_message_id')
            if isinstance(bot_message, dict)
            else getattr(bot_message, 'resp_message_id', None)
        )
        if not message_id or message_id not in self._stream_ctx:
            # 非流式场景（如群聊不支持流式），累积文本后一次性回复
            if chunk_text:
                self._fallback_text[message_id] = self._fallback_text.get(message_id, '') + chunk_text
                self._fallback_text_ts[message_id] = time.time()
            if is_final:
                full_text = self._fallback_text.pop(message_id, '')
                if full_text:
                    fallback_msg = platform_message.MessageChain([platform_message.Plain(text=full_text)])
                    await self.reply_message(message_source, fallback_msg, quote_origin)
            return
        ctx = self._stream_ctx[message_id]
        # 累积文本
        if chunk_text:
            ctx['accumulated_text'] += chunk_text
        # 未启动会话时，等第一个有内容的 chunk 来建立会话
        if not ctx['session_started']:
            if not ctx['accumulated_text']:
                return
            # 用第一个 chunk 的文本建立会话（不发 "..." 避免污染前缀）
            ctx['session_started'] = True
        # 发送内容 = 全量累积文本
        # QQ API 的 replace 模式不允许修改已下发前缀，所以：
        # - 首次：发送全部文本，建立会话
        # - 后续：只能发送新增部分（append 行为）
        content_to_send = ctx['accumulated_text'][ctx['sent_length'] :]
        if not content_to_send and not is_final:
            return
        input_state = 10 if is_final else 1
        # Rate limiting: skip non-final updates if last update was <0.5s ago
        now = time.time()
        if not is_final and (now - ctx['last_update_ts']) < 0.5:
            return
        ctx['last_update_ts'] = now
        try:
            resp = await self.bot.send_stream_msg(
                user_openid=ctx['user_openid'],
                content=content_to_send,
                event_id=ctx['msg_id'],
                msg_id=ctx['msg_id'],
                msg_seq=ctx['msg_seq'],
                index=ctx['index'],
                stream_msg_id=ctx['stream_msg_id'],
                input_state=input_state,
            )
            if resp and isinstance(resp, dict):
                new_stream_id = resp.get('id')
                if new_stream_id:
                    ctx['stream_msg_id'] = new_stream_id
            ctx['sent_length'] = len(ctx['accumulated_text'])
            ctx['index'] += 1
            await self.logger.debug(
                f'[QQ Official] 流式 chunk 已发送, index={ctx["index"]}, '
                f'sent_len={ctx["sent_length"]}, is_final={is_final}'
            )
        except Exception as e:
            await self.logger.error(f'Failed to send stream message: {e}')
        if is_final:
            self._stream_ctx.pop(message_id, None)
    def unregister_listener(
        self,
--- a/src/langbot/pkg/platform/sources/qqofficial.yaml
+++ b/src/langbot/pkg/platform/sources/qqofficial.yaml
@@ -7,9 +7,9 @@ metadata:
    zh_Hans: QQ 官方 API
    zh_Hant: QQ 官方 API
  description:
-    en_US: QQ Official API (Webhook)
+    en_US: QQ Official API (Webhook / WebSocket)
-    zh_Hans: QQ 官方 API (Webhook)，需要公网地址以接收消息推送，请查看文档了解使用方式
+    zh_Hans: QQ 官方 API，支持 Webhook 和 WebSocket 两种连接模式
-    zh_Hant: QQ 官方 API (Webhook)，需要公網地址以接收訊息推送，請查看文件了解使用方式
+    zh_Hant: QQ 官方 API，支援 Webhook 和 WebSocket 兩種連線模式
  icon: qqofficial.svg
 spec:
  categories:
@@ -19,18 +19,6 @@ spec:
    en: https://link.langbot.app/en/platforms/qqofficial
    ja: https://link.langbot.app/ja/platforms/qqofficial
  config:
    - name: webhook_url
      label:
        en_US: Webhook Callback URL
        zh_Hans: Webhook 回调地址
        zh_Hant: Webhook 回調地址
      description:
        en_US: Copy this URL and paste it into your QQ Official API webhook configuration
        zh_Hans: 复制此地址并粘贴到 QQ 官方 API 的 Webhook 配置中
        zh_Hant: 複製此地址並貼到 QQ 官方 API 的 Webhook 設定中
      type: webhook-url
      required: false
      default: ""
    - name: appid
      label:
        en_US: App ID
@@ -55,6 +43,46 @@ spec:
      type: string
      required: true
      default: ""
    - name: enable-webhook
      label:
        en_US: Enable Webhook Mode
        zh_Hans: 启用Webhook模式
        zh_Hant: 啟用 Webhook 模式
      description:
        en_US: If enabled, the bot will use webhook mode to receive messages. Otherwise, it will use WebSocket 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 WebSocket 模式
        zh_Hant: 如果啟用，機器人將使用 Webhook 模式接收訊息。否則，將使用 WebSocket 模式
      type: boolean
      required: true
      default: false
    - name: enable-stream-reply
      label:
        en_US: Enable Stream Reply Mode
        zh_Hans: 启用流式回复模式
        zh_Hant: 啟用串流回覆模式
      description:
        en_US: If enabled, the bot will use streaming mode to reply messages (C2C only)
        zh_Hans: 如果启用，机器人将使用流式方式回复消息（仅私聊）
        zh_Hant: 如果啟用，機器人將使用串流方式回覆訊息（僅私聊）
      type: boolean
      required: true
      default: false
    - name: webhook_url
      label:
        en_US: Webhook Callback URL
        zh_Hans: Webhook 回调地址
        zh_Hant: Webhook 回調地址
      description:
        en_US: Copy this URL and paste it into your QQ Official API webhook configuration
        zh_Hans: 复制此地址并粘贴到 QQ 官方 API 的 Webhook 配置中
        zh_Hant: 複製此地址並貼到 QQ 官方 API 的 Webhook 設定中
      type: webhook-url
      required: false
      default: ""
      show_if:
        field: enable-webhook
        operator: eq
        value: true
 execution:
  python:
    path: ./qqofficial.py
--- a/src/langbot/pkg/platform/sources/web_page_bot.yaml
+++ b/src/langbot/pkg/platform/sources/web_page_bot.yaml
@@ -0,0 +1,177 @@
 apiVersion: v1
 kind: MessagePlatformAdapter
 metadata:
  name: web_page_bot
  label:
    en_US: "Page Bot"
    zh_Hans: "页面机器人"
    zh_Hant: "頁面機器人"
    ja_JP: "ページボット"
    th_TH: "บอทหน้าเว็บ"
    vi_VN: "Bot trang web"
    es_ES: "Bot de página"
  description:
    en_US: "Embed a chat widget on any website with a simple script tag"
    zh_Hans: "通过一行脚本标签将聊天组件嵌入到任何网站"
    zh_Hant: "透過一行腳本標籤將聊天元件嵌入到任何網站"
    ja_JP: "シンプルなスクリプトタグで任意のウェブサイトにチャットウィジェットを埋め込みます"
    th_TH: "ฝังวิดเจ็ตแชทในเว็บไซต์ใดก็ได้ด้วยแท็กสคริปต์"
    vi_VN: "Nhúng widget trò chuyện vào bất kỳ trang web nào bằng thẻ script"
    es_ES: "Incrusta un widget de chat en cualquier sitio web con una etiqueta de script"
  icon: "webpage.webp"
 spec:
  categories:
    - popular
  config:
    - name: title
      label:
        en_US: Widget Title
        zh_Hans: 组件标题
        zh_Hant: 元件標題
        ja_JP: ウィジェットタイトル
        th_TH: ชื่อวิดเจ็ต
        vi_VN: Tiêu đề widget
        es_ES: Título del widget
      description:
        en_US: The title displayed in the chat widget header
        zh_Hans: 显示在聊天组件顶部的标题
        zh_Hant: 顯示在聊天元件頂部的標題
        ja_JP: チャットウィジェットのヘッダーに表示されるタイトル
        th_TH: ชื่อที่แสดงในส่วนหัวของวิดเจ็ตแชท
        vi_VN: Tiêu đề hiển thị trong đầu widget trò chuyện
        es_ES: El título que se muestra en el encabezado del widget de chat
      type: string
      required: false
      default: "LangBot"
    - name: bubble_icon
      label:
        en_US: Bubble Icon
        zh_Hans: 气泡图标
        zh_Hant: 氣泡圖示
        ja_JP: バブルアイコン
        th_TH: ไอคอนบับเบิล
        vi_VN: Biểu tượng bong bóng
        es_ES: Icono de burbuja
        ru_RU: Иконка пузырька
      description:
        en_US: "Icon displayed on the floating chat bubble"
        zh_Hans: "浮动聊天气泡上显示的图标"
      type: select
      required: false
      default: "logo"
      options:
        - name: "logo"
          label:
            en_US: "LangBot Logo"
            zh_Hans: "LangBot 图标"
        - name: "chat"
          label:
            en_US: "Chat Bubble"
            zh_Hans: "聊天气泡"
        - name: "robot"
          label:
            en_US: "Robot"
            zh_Hans: "机器人"
        - name: "headset"
          label:
            en_US: "Headset"
            zh_Hans: "客服耳机"
        - name: "sparkle"
          label:
            en_US: "Sparkle"
            zh_Hans: "星光"
        - name: "message"
          label:
            en_US: "Message"
            zh_Hans: "消息"
    - name: language
      label:
        en_US: Widget Language
        zh_Hans: 组件语言
        zh_Hant: 元件語言
        ja_JP: ウィジェット言語
        th_TH: ภาษาวิดเจ็ต
        vi_VN: Ngôn ngữ widget
        es_ES: Idioma del widget
        ru_RU: Язык виджета
      description:
        en_US: "Display language of the chat widget"
        zh_Hans: "聊天组件的显示语言"
        zh_Hant: "聊天元件的顯示語言"
        ja_JP: "チャットウィジェットの表示言語"
        th_TH: "ภาษาแสดงผลของวิดเจ็ตแชท"
        vi_VN: "Ngôn ngữ hiển thị của widget trò chuyện"
        es_ES: "Idioma de visualización del widget de chat"
        ru_RU: "Язык отображения виджета чата"
      type: select
      required: false
      default: "en_US"
      options:
        - name: "en_US"
          label:
            en_US: "English"
        - name: "zh_Hans"
          label:
            en_US: "简体中文"
        - name: "zh_Hant"
          label:
            en_US: "繁體中文"
        - name: "ja_JP"
          label:
            en_US: "日本語"
        - name: "es_ES"
          label:
            en_US: "Español"
        - name: "ru_RU"
          label:
            en_US: "Русский"
        - name: "th_TH"
          label:
            en_US: "ไทย"
        - name: "vi_VN"
          label:
            en_US: "Tiếng Việt"
    - name: embed_code
      label:
        en_US: Embed Code
        zh_Hans: 嵌入代码
        zh_Hant: 嵌入代碼
        ja_JP: 埋め込みコード
        th_TH: โค้ดฝังตัว
        vi_VN: Mã nhúng
        es_ES: Código de incrustación
      description:
        en_US: "Copy this code and paste it into your website HTML. The code will be generated after saving."
        zh_Hans: "复制此代码并粘贴到你的网站 HTML 中。保存后将自动生成。"
        zh_Hant: "複製此代碼並貼到你的網站 HTML 中。儲存後將自動生成。"
        ja_JP: "このコードをコピーしてウェブサイトのHTMLに貼り付けてください。保存後に自動生成されます。"
        th_TH: "คัดลอกโค้ดนี้และวางในHTML ของเว็บไซต์ของคุณ จะสร้างอัตโนมัติหลังจากบันทึก"
        vi_VN: "Sao chép mã này và dán vào HTML trang web của bạn. Mã sẽ được tạo tự động sau khi lưu."
        es_ES: "Copia este código y pégalo en el HTML de tu sitio web. El código se generará después de guardar."
      type: embed-code
      required: false
      default: ""
    - name: turnstile_site_key
      label:
        en_US: Turnstile Site Key
        zh_Hans: Turnstile 站点密钥
      description:
        en_US: "Cloudflare Turnstile site key for bot protection. Get it from the Cloudflare dashboard (Turnstile > Add Site). Leave empty to disable."
        zh_Hans: "Cloudflare Turnstile 站点密钥，用于防止机器人滥用。在 Cloudflare 控制台（Turnstile > 添加站点）中获取。留空则不启用。"
      type: string
      required: false
      default: ""
    - name: turnstile_secret_key
      label:
        en_US: Turnstile Secret Key
        zh_Hans: Turnstile 服务端密钥
      description:
        en_US: "Cloudflare Turnstile secret key for server-side token verification. Found alongside the site key in the Cloudflare dashboard. Required if site key is set."
        zh_Hans: "Cloudflare Turnstile 服务端密钥，用于服务端验证令牌。与站点密钥一起在 Cloudflare 控制台中获取。设置了站点密钥时必填。"
      type: string
      required: false
      default: ""
 execution:
  python:
    path: "web_page_bot_adapter.py"
    attr: "WebPageBotAdapter"
--- a/src/langbot/pkg/platform/sources/web_page_bot_adapter.py
+++ b/src/langbot/pkg/platform/sources/web_page_bot_adapter.py
@@ -0,0 +1,94 @@
 """Web Page Bot adapter - lightweight adapter for embeddable chat widget"""
 import typing
 import pydantic
 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
 import langbot_plugin.api.definition.abstract.platform.event_logger as abstract_platform_logger
 class WebPageBotAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter):
    """Lightweight adapter for the embeddable page bot.
    This adapter does not handle messages itself. The actual WebSocket
    communication is handled by the singleton websocket_proxy_bot.
    This adapter stores event listeners so that RuntimeBot can register
    its handlers, which are then called by the websocket adapter when
    a message arrives for this bot's pipeline.
    Message sending/replying is delegated to the websocket_proxy_bot's
    adapter so that replies are actually delivered over the WebSocket
    connection while the dashboard correctly shows this adapter's name.
    """
    listeners: dict = pydantic.Field(default_factory=dict, exclude=True)
    _ws_adapter: typing.Any = None
    model_config = pydantic.ConfigDict(arbitrary_types_allowed=True)
    def __init__(self, config: dict, logger: abstract_platform_logger.AbstractEventLogger, **kwargs):
        super().__init__(config=config, logger=logger, **kwargs)
    def set_ws_adapter(self, ws_adapter) -> None:
        """Set the underlying WebSocket adapter used for actual message delivery."""
        object.__setattr__(self, '_ws_adapter', ws_adapter)
    async def send_message(
        self,
        target_type: str,
        target_id: str,
        message: platform_message.MessageChain,
    ) -> dict:
        if self._ws_adapter is not None:
            return await self._ws_adapter.send_message(target_type, target_id, message)
        return {}
    async def reply_message(
        self,
        message_source: platform_events.MessageEvent,
        message: platform_message.MessageChain,
        quote_origin: bool = False,
    ) -> dict:
        if self._ws_adapter is not None:
            return await self._ws_adapter.reply_message(message_source, message, quote_origin)
        return {}
    async def reply_message_chunk(
        self,
        message_source: platform_events.MessageEvent,
        bot_message,
        message: platform_message.MessageChain,
        quote_origin: bool = False,
        is_final: bool = False,
    ) -> dict:
        if self._ws_adapter is not None:
            return await self._ws_adapter.reply_message_chunk(
                message_source, bot_message, message, quote_origin, is_final
            )
        return {}
    def register_listener(
        self,
        event_type: typing.Type[platform_events.Event],
        func: typing.Callable,
    ):
        self.listeners[event_type] = func
    def unregister_listener(
        self,
        event_type: typing.Type[platform_events.Event],
        func: typing.Callable,
    ):
        self.listeners.pop(event_type, None)
    async def is_muted(self, group_id: int) -> bool:
        return False
    async def run_async(self):
        pass
    async def kill(self):
        pass
--- a/src/langbot/pkg/platform/sources/webpage.webp
+++ b/src/langbot/pkg/platform/sources/webpage.webp
--- a/src/langbot/pkg/platform/sources/websocket_adapter.py
+++ b/src/langbot/pkg/platform/sources/websocket_adapter.py
@@ -312,7 +312,7 @@ class WebSocketAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter)
    async def _process_image_components(self, message_chain_obj: list):
        """
-        处理消息链中的图片组件，将path转换为base64
+        处理消息链中的图片和文件组件，将path转换为base64
        Args:
            message_chain_obj: 消息链对象列表
@@ -322,16 +322,18 @@ class WebSocketAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter)
        storage_mgr = self.ap.storage_mgr
        for component in message_chain_obj:
-            if component.get('type') == 'Image' and component.get('path'):
+            comp_type = component.get('type', '')
-                try:
+            comp_path = component.get('path', '')
                    # 从storage读取文件
                    file_content = await storage_mgr.storage_provider.load(component['path'])
-                    # 转换为base64
+            if not comp_path:
                continue
            if comp_type == 'Image':
                try:
                    file_content = await storage_mgr.storage_provider.load(comp_path)
                    base64_str = base64.b64encode(file_content).decode('utf-8')
-                    # 添加data URI前缀（根据文件扩展名判断MIME类型）
+                    file_key = comp_path
                    file_key = component['path']
                    if file_key.lower().endswith(('.jpg', '.jpeg')):
                        mime_type = 'image/jpeg'
                    elif file_key.lower().endswith('.png'):
@@ -341,19 +343,19 @@ class WebSocketAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter)
                    elif file_key.lower().endswith('.webp'):
                        mime_type = 'image/webp'
                    else:
-                        mime_type = 'image/png'  # 默认
+                        mime_type = 'image/png'
                    component['base64'] = f'data:{mime_type};base64,{base64_str}'
-                    await storage_mgr.storage_provider.delete(component['path'])
+                    await storage_mgr.storage_provider.delete(comp_path)
                    component['path'] = ''
                    # 保留path字段用于后端处理，前端使用base64显示
                except Exception as e:
-                    await self.logger.error(f'加载图片文件失败 {component["path"]}: {e}')
+                    await self.logger.error(f'Failed to load image file {comp_path}: {e}')
    async def handle_websocket_message(
        self,
        connection: WebSocketConnection,
        message_data: dict,
        owner_bot=None,
    ):
        """
        处理从WebSocket接收的消息
@@ -366,6 +368,8 @@ class WebSocketAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter)
            message_data: 消息数据，包含:
                - message: 消息链
                - stream: 是否启用流式输出 (可选，默认True)
            owner_bot: Optional RuntimeBot that owns this pipeline (e.g. a web_page_bot).
                       When provided, its identity is used for logging and session tracking.
        """
        pipeline_uuid = connection.pipeline_uuid
        session_type = connection.session_type
@@ -435,12 +439,26 @@ class WebSocketAdapter(abstract_platform_adapter.AbstractMessagePlatformAdapter)
                sender=sender, message_chain=message_chain, time=datetime.now().timestamp()
            )
-        # 设置流水线UUID
+        # 设置流水线UUID (proxy bot always needs it for reply_message routing)
        self.ap.platform_mgr.websocket_proxy_bot.bot_entity.use_pipeline_uuid = pipeline_uuid
        if owner_bot is not None:
            owner_bot.bot_entity.use_pipeline_uuid = pipeline_uuid
-        # 异步触发事件处理（不等待结果）
+        # 异步触发事件处理
-        if event.__class__ in self.listeners:
+        # Use owner_bot's listeners if available, otherwise fall back to proxy bot
-            asyncio.create_task(self.listeners[event.__class__](event, self))
+        listeners = (
            owner_bot.adapter.listeners
            if (owner_bot and hasattr(owner_bot.adapter, 'listeners') and owner_bot.adapter.listeners)
            else self.listeners
        )
        # Pass owner_bot's adapter so that downstream logging / dashboard
        # attributes the message to the correct bot adapter name.
        # Wire the ws adapter into the owner so replies are actually delivered.
        if owner_bot and hasattr(owner_bot.adapter, 'set_ws_adapter'):
            owner_bot.adapter.set_ws_adapter(self)
        callback_adapter = owner_bot.adapter if (owner_bot and hasattr(owner_bot, 'adapter')) else self
        if event.__class__ in listeners:
            asyncio.create_task(listeners[event.__class__](event, callback_adapter))
    def get_websocket_messages(self, pipeline_uuid: str, session_type: str) -> list[dict]:
        """获取消息历史"""
--- a/src/langbot/pkg/platform/sources/wecombot.yaml
+++ b/src/langbot/pkg/platform/sources/wecombot.yaml
@@ -19,6 +19,18 @@ spec:
    en: https://link.langbot.app/en/platforms/wecombot
    ja: https://link.langbot.app/ja/platforms/wecombot
  config:
    - name: one-click-create
      label:
        en_US: One-Click Create Bot
        zh_Hans: 一键创建机器人
        zh_Hant: 一鍵建立機器人
      description:
        en_US: "Scan QR code with WeCom to automatically create a bot and fill in BotId and Secret. Note: Robot Name needs to be filled in manually."
        zh_Hans: "使用企业微信扫码自动创建机器人并填写 BotId 和 Secret。注意：机器人名称需手动填写。"
        zh_Hant: "使用企業微信掃碼自動建立機器人並填寫 BotId 和 Secret。注意：機器人名稱需手動填寫。"
      type: qr-code-login
      login_platform: wecombot
      required: false
    - name: BotId
      label:
        en_US: BotId
--- a/src/langbot/pkg/plugin/connector.py
+++ b/src/langbot/pkg/plugin/connector.py
@@ -11,6 +11,7 @@ import os
 import sys
 import httpx
 import sqlalchemy
 import yaml
 from async_lru import alru_cache
 from langbot_plugin.api.entities.builtin.pipeline.query import provider_session
@@ -195,40 +196,110 @@ class PluginRuntimeConnector:
        return await self.handler.ping()
-    def _extract_deps_metadata(
+    def _inspect_plugin_package(
        self,
        file_bytes: bytes,
        task_context: taskmgr.TaskContext | None,
-    ):
+    ) -> tuple[str | None, str | None]:
-        """Extract dependency count from requirements.txt inside plugin zip."""
+        """Extract plugin identity and dependency metadata from a plugin package."""
-        if task_context is None:
+        plugin_author = None
-            return
+        plugin_name = None
        try:
            with zipfile.ZipFile(io.BytesIO(file_bytes)) as zf:
-                for name in zf.namelist():
+                try:
-                    if name.endswith('requirements.txt'):
+                    manifest = yaml.safe_load(zf.read('manifest.yaml').decode('utf-8', errors='ignore')) or {}
-                        content = zf.read(name).decode('utf-8', errors='ignore')
+                    metadata = manifest.get('metadata', {})
-                        deps = [
+                    plugin_author = metadata.get('author')
-                            line.strip()
+                    plugin_name = metadata.get('name')
-                            for line in content.splitlines()
+                except Exception:
-                            if line.strip() and not line.strip().startswith('#')
+                    pass
-                        ]
+
-                        task_context.metadata['deps_total'] = len(deps)
+                if task_context is not None:
-                        task_context.metadata['deps_list'] = deps
+                    for name in zf.namelist():
-                        break
+                        if name.endswith('requirements.txt'):
                            content = zf.read(name).decode('utf-8', errors='ignore')
                            deps = [
                                line.strip()
                                for line in content.splitlines()
                                if line.strip() and not line.strip().startswith('#')
                            ]
                            task_context.metadata['deps_total'] = len(deps)
                            task_context.metadata['deps_list'] = deps
                            break
        except Exception:
            pass
        return plugin_author, plugin_name
    def _build_plugin_startup_failure_message(
        self,
        plugin_author: str,
        plugin_name: str,
        task_context: taskmgr.TaskContext | None,
    ) -> str:
        dep_hint = ''
        if task_context is not None:
            current_dep = task_context.metadata.get('current_dep')
            if current_dep:
                dep_hint = f' Last dependency: {current_dep}.'
        return (
            f'Plugin {plugin_author}/{plugin_name} failed to start after installation. '
            f'Dependency installation or plugin initialization may have failed.{dep_hint} '
            f'Please check the plugin requirements and runtime logs.'
        )
    async def _wait_for_installed_plugin_ready(
        self,
        plugin_author: str | None,
        plugin_name: str | None,
        task_context: taskmgr.TaskContext | None,
        timeout: float = 30,
    ):
        """Wait until the installed plugin is registered by the runtime.
        The plugin runtime launches plugins asynchronously. If dependency installation
        fails, the plugin process exits before registration; without this check the
        install task can incorrectly finish successfully.
        """
        if not plugin_author or not plugin_name:
            return
        deadline = time.time() + timeout
        last_error: Exception | None = None
        while time.time() < deadline:
            try:
                plugin = await self.get_plugin_info(plugin_author, plugin_name)
                if plugin is not None:
                    status = plugin.get('status')
                    if status == 'initialized':
                        return
            except Exception as e:
                last_error = e
            await asyncio.sleep(0.5)
        message = self._build_plugin_startup_failure_message(plugin_author, plugin_name, task_context)
        if last_error is not None:
            message = f'{message} Last runtime error: {last_error}'
        raise RuntimeError(message)
    async def install_plugin(
        self,
        install_source: PluginInstallSource,
        install_info: dict[str, Any],
        task_context: taskmgr.TaskContext | None = None,
    ):
        plugin_author = install_info.get('plugin_author')
        plugin_name = install_info.get('plugin_name')
        if install_source == PluginInstallSource.LOCAL:
            # transfer file before install
            file_bytes = install_info['plugin_file']
-            self._extract_deps_metadata(file_bytes, task_context)
+            plugin_author, plugin_name = self._inspect_plugin_package(file_bytes, task_context)
            if task_context is not None and plugin_author and plugin_name:
                task_context.metadata['plugin_name'] = f'{plugin_author}/{plugin_name}'
            file_key = await self.handler.send_file(file_bytes, 'lbpkg')
            install_info['plugin_file_key'] = file_key
            del install_info['plugin_file']
@@ -265,7 +336,9 @@ class PluginRuntimeConnector:
                                task_context.metadata['download_speed'] = downloaded / elapsed if elapsed > 0 else 0
                    file_bytes = b''.join(chunks)
-                    self._extract_deps_metadata(file_bytes, task_context)
+                    plugin_author, plugin_name = self._inspect_plugin_package(file_bytes, task_context)
                    if task_context is not None and plugin_author and plugin_name:
                        task_context.metadata['plugin_name'] = f'{plugin_author}/{plugin_name}'
                    file_key = await self.handler.send_file(file_bytes, 'lbpkg')
                    install_info['plugin_file_key'] = file_key
                    self.ap.logger.info(f'Transfered file {file_key} to plugin runtime')
@@ -289,6 +362,8 @@ class PluginRuntimeConnector:
            if metadata is not None and task_context is not None:
                task_context.metadata.update(metadata)
        await self._wait_for_installed_plugin_ready(plugin_author, plugin_name, task_context)
    async def upgrade_plugin(
        self,
        plugin_author: str,
@@ -431,6 +506,17 @@ class PluginRuntimeConnector:
    async def get_plugin_assets(self, plugin_author: str, plugin_name: str, filepath: str) -> dict[str, Any]:
        return await self.handler.get_plugin_assets(plugin_author, plugin_name, filepath)
    async def handle_page_api(
        self,
        plugin_author: str,
        plugin_name: str,
        page_id: str,
        endpoint: str,
        method: str,
        body: Any = None,
    ) -> dict[str, Any]:
        return await self.handler.handle_page_api(plugin_author, plugin_name, page_id, endpoint, method, body)
    async def get_debug_info(self) -> dict[str, Any]:
        """Get debug information including debug key and WS URL"""
        if not self.is_enable_plugin:
--- a/src/langbot/pkg/plugin/handler.py
+++ b/src/langbot/pkg/plugin/handler.py
@@ -367,6 +367,22 @@ class RuntimeConnectionHandler(handler.Handler):
            owner_type = data['owner_type']
            owner = data['owner']
            value = base64.b64decode(data['value_base64'])
            max_value_bytes = (
                self.ap.instance_config.data.get('plugin', {})
                .get('binary_storage', {})
                .get(
                    'max_value_bytes',
                    10 * 1024 * 1024,
                )
            )
            try:
                max_value_bytes = int(max_value_bytes)
            except (TypeError, ValueError):
                max_value_bytes = 10 * 1024 * 1024
            if max_value_bytes >= 0 and len(value) > max_value_bytes:
                return handler.ActionResponse.error(
                    message=f'Binary storage value exceeds limit ({len(value)} > {max_value_bytes} bytes)',
                )
            result = await self.ap.persistence_mgr.execute_async(
                sqlalchemy.select(persistence_bstorage.BinaryStorage)
@@ -939,6 +955,11 @@ class RuntimeConnectionHandler(handler.Handler):
            timeout=20,
        )
        asset_file_key = result['file_file_key']
        if not asset_file_key:
            return {
                'asset_base64': '',
                'mime_type': '',
            }
        mime_type = result['mime_type']
        asset_bytes = await self.read_local_file(asset_file_key)
        await self.delete_local_file(asset_file_key)
@@ -947,6 +968,30 @@ class RuntimeConnectionHandler(handler.Handler):
            'mime_type': mime_type,
        }
    async def handle_page_api(
        self,
        plugin_author: str,
        plugin_name: str,
        page_id: str,
        endpoint: str,
        method: str,
        body: Any = None,
    ) -> dict[str, Any]:
        """Forward a page API call to the plugin via runtime."""
        result = await self.call_action(
            LangBotToRuntimeAction.PAGE_API,
            {
                'plugin_author': plugin_author,
                'plugin_name': plugin_name,
                'page_id': page_id,
                'endpoint': endpoint,
                'method': method,
                'body': body,
            },
            timeout=30,
        )
        return result
    async def cleanup_plugin_data(self, plugin_author: str, plugin_name: str) -> None:
        """Cleanup plugin settings and binary storage"""
        # Delete plugin settings
--- a/src/langbot/pkg/provider/modelmgr/modelmgr.py
+++ b/src/langbot/pkg/provider/modelmgr/modelmgr.py
@@ -4,7 +4,6 @@ import sqlalchemy
 import traceback
 from . import requester
 from .requesters import litellmchat
 from ...core import app
 from ...discover import engine
 from . import token
@@ -43,13 +42,6 @@ class ModelManager:
        requester_dict: dict[str, type[requester.ProviderAPIRequester]] = {}
        for component in self.requester_components:
            # Skip components that use litellm_provider (they will use litellmchat.py instead)
            if component.spec.get('litellm_provider'):
                self.ap.logger.debug(
                    f'Skipping Python class loading for {component.metadata.name} '
                    f'(uses litellm_provider={component.spec.get("litellm_provider")})'
                )
                continue
            requester_dict[component.metadata.name] = component.get_python_component_class()
        self.requester_dict = requester_dict
@@ -268,34 +260,13 @@ class ModelManager:
        else:
            provider_entity = provider_info
-        # Get requester manifest to check for litellm_provider
+        if provider_entity.requester not in self.requester_dict:
-        requester_manifest = self.get_available_requester_manifest_by_name(provider_entity.requester)
+            raise provider_errors.RequesterNotFoundError(provider_entity.requester)
        # Build config from base_url
        config = {'base_url': provider_entity.base_url}
        # Check if requester manifest specifies litellm_provider
        if requester_manifest and requester_manifest.spec.get('litellm_provider'):
            # Use unified LiteLLMRequester with provider prefix
            # Map litellm_provider (YAML spec) to custom_llm_provider (config)
            config['custom_llm_provider'] = requester_manifest.spec['litellm_provider']
            requester_inst = litellmchat.LiteLLMRequester(
                ap=self.ap,
                config=config,
            )
            self.ap.logger.debug(
                f'Using LiteLLMRequester for {provider_entity.requester} '
                f'with custom_llm_provider={config["custom_llm_provider"]}'
            )
        else:
            # Use original requester class (for backward compatibility)
            if provider_entity.requester not in self.requester_dict:
                raise provider_errors.RequesterNotFoundError(provider_entity.requester)
            requester_inst = self.requester_dict[provider_entity.requester](
                ap=self.ap,
                config=config,
            )
        requester_inst = self.requester_dict[provider_entity.requester](
            ap=self.ap,
            config={'base_url': provider_entity.base_url},
        )
        await requester_inst.initialize()
        token_mgr = token.TokenManager(name=provider_entity.uuid, tokens=provider_entity.api_keys or [])
--- a/src/langbot/pkg/provider/modelmgr/requester.py
+++ b/src/langbot/pkg/provider/modelmgr/requester.py
@@ -67,8 +67,8 @@ class RuntimeProvider:
            if isinstance(result, tuple):
                msg, usage_info = result
                if usage_info:
-                    input_tokens = usage_info.get('prompt_tokens', 0)
+                    input_tokens = usage_info.get('input_tokens', 0)
-                    output_tokens = usage_info.get('completion_tokens', 0)
+                    output_tokens = usage_info.get('output_tokens', 0)
                return msg
            else:
                return result
@@ -128,6 +128,7 @@ class RuntimeProvider:
        start_time = time.time()
        status = 'success'
        error_message = None
        # Note: Stream doesn't easily provide token counts, set to 0
        input_tokens = 0
        output_tokens = 0
@@ -142,15 +143,6 @@ class RuntimeProvider:
                remove_think=remove_think,
            ):
                yield chunk
            # Extract usage from stream if available (stored by LiteLLM requester)
            if query:
                if query.variables is None:
                    query.variables = {}
                if '_stream_usage' in query.variables:
                    usage_info = query.variables['_stream_usage']
                    input_tokens = usage_info.get('prompt_tokens', 0)
                    output_tokens = usage_info.get('completion_tokens', 0)
                    del query.variables['_stream_usage']
        except Exception as e:
            status = 'error'
            error_message = str(e)
@@ -348,6 +340,7 @@ class ProviderAPIRequester(metaclass=abc.ABCMeta):
    """Provider API请求器"""
    name: str = None
    init_api_key: str = 'langbot-init-placeholder'
    ap: app.Application
--- a/src/langbot/pkg/provider/modelmgr/requesters/302aichatcmpl.py
+++ b/src/langbot/pkg/provider/modelmgr/requesters/302aichatcmpl.py
@@ -0,0 +1,17 @@
 from __future__ import annotations
 import typing
 import openai
 from . import chatcmpl
 class AI302ChatCompletions(chatcmpl.OpenAIChatCompletions):
    """302.AI ChatCompletion API 请求器"""
    client: openai.AsyncClient
    default_config: dict[str, typing.Any] = {
        'base_url': 'https://api.302.ai/v1',
        'timeout': 120,
    }
--- a/src/langbot/pkg/provider/modelmgr/requesters/302aichatcmpl.yaml
+++ b/src/langbot/pkg/provider/modelmgr/requesters/302aichatcmpl.yaml
@@ -7,7 +7,6 @@ metadata:
    zh_Hans: 302.AI
  icon: 302ai.png
 spec:
  litellm_provider: openai
  config:
  - name: base_url
    label:
--- a/src/langbot/pkg/provider/modelmgr/requesters/anthropicmsgs.py
+++ b/src/langbot/pkg/provider/modelmgr/requesters/anthropicmsgs.py
@@ -0,0 +1,370 @@
 from __future__ import annotations
 import typing
 import json
 import platform
 import socket
 import anthropic
 import httpx
 from .. import errors, requester
 from ....utils import image
 import langbot_plugin.api.entities.builtin.resource.tool as resource_tool
 import langbot_plugin.api.entities.builtin.pipeline.query as pipeline_query
 import langbot_plugin.api.entities.builtin.provider.message as provider_message
 class AnthropicMessages(requester.ProviderAPIRequester):
    """Anthropic Messages API 请求器"""
    client: anthropic.AsyncAnthropic
    default_config: dict[str, typing.Any] = {
        'base_url': 'https://api.anthropic.com',
        'timeout': 120,
    }
    async def initialize(self):
        # 兼容 Windows 缺失 TCP_KEEPINTVL 和 TCP_KEEPCNT 的问题
        if platform.system() == 'Windows':
            if not hasattr(socket, 'TCP_KEEPINTVL'):
                socket.TCP_KEEPINTVL = 0
            if not hasattr(socket, 'TCP_KEEPCNT'):
                socket.TCP_KEEPCNT = 0
        httpx_client = anthropic._base_client.AsyncHttpxClientWrapper(
            base_url=self.requester_cfg['base_url'],
            # cast to a valid type because mypy doesn't understand our type narrowing
            timeout=typing.cast(httpx.Timeout, self.requester_cfg['timeout']),
            limits=anthropic._constants.DEFAULT_CONNECTION_LIMITS,
            follow_redirects=True,
            trust_env=True,
        )
        self.client = anthropic.AsyncAnthropic(
            api_key='',
            http_client=httpx_client,
            base_url=self.requester_cfg['base_url'],
        )
    async def invoke_llm(
        self,
        query: pipeline_query.Query,
        model: requester.RuntimeLLMModel,
        messages: typing.List[provider_message.Message],
        funcs: typing.List[resource_tool.LLMTool] = None,
        extra_args: dict[str, typing.Any] = {},
        remove_think: bool = False,
    ) -> provider_message.Message:
        self.client.api_key = model.provider.token_mgr.get_token()
        args = extra_args.copy()
        args['model'] = model.model_entity.name
        # 处理消息
        # system
        system_role_message = None
        for i, m in enumerate(messages):
            if m.role == 'system':
                system_role_message = m
                break
        if system_role_message:
            messages.pop(i)
        if isinstance(system_role_message, provider_message.Message) and isinstance(system_role_message.content, str):
            args['system'] = system_role_message.content
        req_messages = []
        for m in messages:
            if m.role == 'tool':
                tool_call_id = m.tool_call_id
                req_messages.append(
                    {
                        'role': 'user',
                        'content': [
                            {
                                'type': 'tool_result',
                                'tool_use_id': tool_call_id,
                                'is_error': False,
                                'content': [{'type': 'text', 'text': m.content}],
                            }
                        ],
                    }
                )
                continue
            msg_dict = m.dict(exclude_none=True)
            if isinstance(m.content, str) and m.content.strip() != '':
                msg_dict['content'] = [{'type': 'text', 'text': m.content}]
            elif isinstance(m.content, list):
                for i, ce in enumerate(m.content):
                    if ce.type == 'image_base64':
                        image_b64, image_format = await image.extract_b64_and_format(ce.image_base64)
                        alter_image_ele = {
                            'type': 'image',
                            'source': {
                                'type': 'base64',
                                'media_type': f'image/{image_format}',
                                'data': image_b64,
                            },
                        }
                        msg_dict['content'][i] = alter_image_ele
            if m.tool_calls:
                for tool_call in m.tool_calls:
                    msg_dict['content'].append(
                        {
                            'type': 'tool_use',
                            'id': tool_call.id,
                            'name': tool_call.function.name,
                            'input': json.loads(tool_call.function.arguments),
                        }
                    )
                del msg_dict['tool_calls']
            req_messages.append(msg_dict)
        args['messages'] = req_messages
        if 'thinking' in args:
            args['thinking'] = {'type': 'enabled', 'budget_tokens': 10000}
        if funcs:
            tools = await self.ap.tool_mgr.generate_tools_for_anthropic(funcs)
            if tools:
                args['tools'] = tools
        try:
            resp = await self.client.messages.create(**args)
            args = {
                'content': '',
                'role': resp.role,
            }
            assert type(resp) is anthropic.types.message.Message
            for block in resp.content:
                if not remove_think and block.type == 'thinking':
                    args['content'] = '<think>\n' + block.thinking + '\n</think>\n' + args['content']
                elif block.type == 'text':
                    args['content'] += block.text
                elif block.type == 'tool_use':
                    assert type(block) is anthropic.types.tool_use_block.ToolUseBlock
                    tool_call = provider_message.ToolCall(
                        id=block.id,
                        type='function',
                        function=provider_message.FunctionCall(name=block.name, arguments=json.dumps(block.input)),
                    )
                    if 'tool_calls' not in args:
                        args['tool_calls'] = []
                    args['tool_calls'].append(tool_call)
            return provider_message.Message(**args)
        except anthropic.AuthenticationError as e:
            raise errors.RequesterError(f'api-key 无效: {e.message}')
        except anthropic.BadRequestError as e:
            raise errors.RequesterError(str(e.message))
        except anthropic.NotFoundError as e:
            if 'model: ' in str(e):
                raise errors.RequesterError(f'模型无效: {e.message}')
            else:
                raise errors.RequesterError(f'请求地址无效: {e.message}')
    async def invoke_llm_stream(
        self,
        query: pipeline_query.Query,
        model: requester.RuntimeLLMModel,
        messages: typing.List[provider_message.Message],
        funcs: typing.List[resource_tool.LLMTool] = None,
        extra_args: dict[str, typing.Any] = {},
        remove_think: bool = False,
    ) -> provider_message.Message:
        self.client.api_key = model.provider.token_mgr.get_token()
        args = extra_args.copy()
        args['model'] = model.model_entity.name
        args['stream'] = True
        # 处理消息
        # system
        system_role_message = None
        for i, m in enumerate(messages):
            if m.role == 'system':
                system_role_message = m
                break
        if system_role_message:
            messages.pop(i)
        if isinstance(system_role_message, provider_message.Message) and isinstance(system_role_message.content, str):
            args['system'] = system_role_message.content
        req_messages = []
        for m in messages:
            if m.role == 'tool':
                tool_call_id = m.tool_call_id
                req_messages.append(
                    {
                        'role': 'user',
                        'content': [
                            {
                                'type': 'tool_result',
                                'tool_use_id': tool_call_id,
                                'is_error': False,  # 暂时直接写false
                                'content': [
                                    {'type': 'text', 'text': m.content}
                                ],  # 这里要是list包裹，应该是多个返回的情况？type类型好像也可以填其他的，暂时只写text
                            }
                        ],
                    }
                )
                continue
            msg_dict = m.dict(exclude_none=True)
            if isinstance(m.content, str) and m.content.strip() != '':
                msg_dict['content'] = [{'type': 'text', 'text': m.content}]
            elif isinstance(m.content, list):
                for i, ce in enumerate(m.content):
                    if ce.type == 'image_base64':
                        image_b64, image_format = await image.extract_b64_and_format(ce.image_base64)
                        alter_image_ele = {
                            'type': 'image',
                            'source': {
                                'type': 'base64',
                                'media_type': f'image/{image_format}',
                                'data': image_b64,
                            },
                        }
                        msg_dict['content'][i] = alter_image_ele
            if isinstance(msg_dict['content'], str) and msg_dict['content'] == '':
                msg_dict['content'] = []  # 这里不知道为什么会莫名有个空导致content为字符
            if m.tool_calls:
                for tool_call in m.tool_calls:
                    msg_dict['content'].append(
                        {
                            'type': 'tool_use',
                            'id': tool_call.id,
                            'name': tool_call.function.name,
                            'input': json.loads(tool_call.function.arguments),
                        }
                    )
                del msg_dict['tool_calls']
            req_messages.append(msg_dict)
        if 'thinking' in args:
            args['thinking'] = {'type': 'enabled', 'budget_tokens': 10000}
        args['messages'] = req_messages
        if funcs:
            tools = await self.ap.tool_mgr.generate_tools_for_anthropic(funcs)
            if tools:
                args['tools'] = tools
        try:
            role = 'assistant'  # 默认角色
            # chunk_idx = 0
            think_started = False
            think_ended = False
            finish_reason = False
            tool_name = ''
            tool_id = ''
            async for chunk in await self.client.messages.create(**args):
                content = ''
                tool_call = {'id': None, 'function': {'name': None, 'arguments': None}, 'type': 'function'}
                if isinstance(
                    chunk, anthropic.types.raw_content_block_start_event.RawContentBlockStartEvent
                ):  # 记录开始
                    if chunk.content_block.type == 'tool_use':
                        if chunk.content_block.name is not None:
                            tool_name = chunk.content_block.name
                        if chunk.content_block.id is not None:
                            tool_id = chunk.content_block.id
                        tool_call['function']['name'] = tool_name
                        tool_call['function']['arguments'] = ''
                        tool_call['id'] = tool_id
                    if not remove_think:
                        if chunk.content_block.type == 'thinking' and not remove_think:
                            think_started = True
                        elif chunk.content_block.type == 'text' and chunk.index != 0 and not remove_think:
                            think_ended = True
                        continue
                elif isinstance(chunk, anthropic.types.raw_content_block_delta_event.RawContentBlockDeltaEvent):
                    if chunk.delta.type == 'thinking_delta':
                        if think_started:
                            think_started = False
                            content = '<think>\n' + chunk.delta.thinking
                        elif remove_think:
                            continue
                        else:
                            content = chunk.delta.thinking
                    elif chunk.delta.type == 'text_delta':
                        if think_ended:
                            think_ended = False
                            content = '\n</think>\n' + chunk.delta.text
                        else:
                            content = chunk.delta.text
                    elif chunk.delta.type == 'input_json_delta':
                        tool_call['function']['arguments'] = chunk.delta.partial_json
                        tool_call['function']['name'] = tool_name
                        tool_call['id'] = tool_id
                elif isinstance(chunk, anthropic.types.raw_content_block_stop_event.RawContentBlockStopEvent):
                    continue  # 记录raw_content_block结束的
                elif isinstance(chunk, anthropic.types.raw_message_delta_event.RawMessageDeltaEvent):
                    if chunk.delta.stop_reason == 'end_turn':
                        finish_reason = True
                elif isinstance(chunk, anthropic.types.raw_message_stop_event.RawMessageStopEvent):
                    continue  # 这个好像是完全结束
                else:
                    # print(chunk)
                    self.ap.logger.debug(f'anthropic chunk: {chunk}')
                    continue
                args = {
                    'content': content,
                    'role': role,
                    'is_final': finish_reason,
                    'tool_calls': None if tool_call['id'] is None else [tool_call],
                }
                # if chunk_idx == 0:
                #     chunk_idx += 1
                #     continue
                # assert type(chunk) is anthropic.types.message.Chunk
                yield provider_message.MessageChunk(**args)
            # return llm_entities.Message(**args)
        except anthropic.AuthenticationError as e:
            raise errors.RequesterError(f'api-key 无效: {e.message}')
        except anthropic.BadRequestError as e:
            raise errors.RequesterError(str(e.message))
        except anthropic.NotFoundError as e:
            if 'model: ' in str(e):
                raise errors.RequesterError(f'模型无效: {e.message}')
            else:
                raise errors.RequesterError(f'请求地址无效: {e.message}')
--- a/src/langbot/pkg/provider/modelmgr/requesters/anthropicmsgs.yaml
+++ b/src/langbot/pkg/provider/modelmgr/requesters/anthropicmsgs.yaml
@@ -7,7 +7,6 @@ metadata:
    zh_Hans: Anthropic
  icon: anthropic.svg
 spec:
  litellm_provider: anthropic
  config:
  - name: base_url
    label:
@@ -25,8 +24,6 @@ spec:
    default: 120
  support_type:
  - llm
  - text-embedding
  - rerank
  provider_category: manufacturer
 execution:
  python:
--- a/src/langbot/pkg/provider/modelmgr/requesters/baidu.svg
+++ b/src/langbot/pkg/provider/modelmgr/requesters/baidu.svg
@@ -1,5 +0,0 @@
 <svg width="60" height="50" viewBox="0 0 60 50" xmlns="http://www.w3.org/2000/svg">
  <rect width="60" height="50" rx="8" fill="#2932E1"/>
  <text x="30" y="28" font-family="Arial, sans-serif" font-size="10" font-weight="bold" fill="white" text-anchor="middle">Baidu</text>
  <text x="30" y="40" font-family="Arial, sans-serif" font-size="8" fill="white" text-anchor="middle">ERNIE</text>
 </svg>
--- a/src/langbot/pkg/provider/modelmgr/requesters/baiduchatcmpl.yaml
+++ b/src/langbot/pkg/provider/modelmgr/requesters/baiduchatcmpl.yaml
@@ -1,30 +0,0 @@
 apiVersion: v1
 kind: LLMAPIRequester
 metadata:
  name: baidu-chat-completions
  label:
    en_US: Baidu ERNIE
    zh_Hans: 百度文心一言
  icon: baidu.svg
 spec:
  litellm_provider: openai
  config:
  - name: base_url
    label:
      en_US: Base URL
      zh_Hans: 基础 URL
    type: string
    required: true
    default: https://aip.baidubce.com/rpc/2.0/ai_custom/v1/wenxinworkshop
  - name: timeout
    label:
      en_US: Timeout
      zh_Hans: 超时时间
    type: integer
    required: true
    default: 120
  support_type:
  - llm
  - text-embedding
  - rerank
  provider_category: manufacturer
--- a/src/langbot/pkg/provider/modelmgr/requesters/bailianchatcmpl.py
+++ b/src/langbot/pkg/provider/modelmgr/requesters/bailianchatcmpl.py
@@ -0,0 +1,242 @@
 from __future__ import annotations
 import typing
 import dashscope
 import openai
 from . import modelscopechatcmpl
 from .. import requester
 import langbot_plugin.api.entities.builtin.resource.tool as resource_tool
 import langbot_plugin.api.entities.builtin.pipeline.query as pipeline_query
 import langbot_plugin.api.entities.builtin.provider.message as provider_message
 class BailianChatCompletions(modelscopechatcmpl.ModelScopeChatCompletions):
    """阿里云百炼大模型平台 ChatCompletion API 请求器"""
    client: openai.AsyncClient
    default_config: dict[str, typing.Any] = {
        'base_url': 'https://dashscope.aliyuncs.com/compatible-mode/v1',
        'timeout': 120,
    }
    async def _closure_stream(
        self,
        query: pipeline_query.Query,
        req_messages: list[dict],
        use_model: requester.RuntimeLLMModel,
        use_funcs: list[resource_tool.LLMTool] = None,
        extra_args: dict[str, typing.Any] = {},
        remove_think: bool = False,
    ) -> provider_message.Message | typing.AsyncGenerator[provider_message.MessageChunk, None]:
        self.client.api_key = use_model.provider.token_mgr.get_token()
        args = {}
        args['model'] = use_model.model_entity.name
        if use_funcs:
            tools = await self.ap.tool_mgr.generate_tools_for_openai(use_funcs)
            if tools:
                args['tools'] = tools
        # 设置此次请求中的messages
        messages = req_messages.copy()
        is_use_dashscope_call = False  # 是否使用阿里原生库调用
        is_enable_multi_model = True  # 是否支持多轮对话
        use_time_num = 0  # 模型已调用次数，防止存在多文件时重复调用
        use_time_ids = []  # 已调用的ID列表
        message_id = 0  # 记录消息序号
        for msg in messages:
            # print(msg)
            if 'content' in msg and isinstance(msg['content'], list):
                for me in msg['content']:
                    if me['type'] == 'image_base64':
                        me['image_url'] = {'url': me['image_base64']}
                        me['type'] = 'image_url'
                        del me['image_base64']
                    elif me['type'] == 'file_url' and '.' in me.get('file_name', ''):
                        # 1. 视频文件推理
                        # https://bailian.console.aliyun.com/?tab=doc#/doc/?type=model&url=2845871
                        file_type = me.get('file_name').lower().split('.')[-1]
                        if file_type in ['mp4', 'avi', 'mkv', 'mov', 'flv', 'wmv']:
                            me['type'] = 'video_url'
                            me['video_url'] = {'url': me['file_url']}
                            del me['file_url']
                            del me['file_name']
                            use_time_num += 1
                            use_time_ids.append(message_id)
                            is_enable_multi_model = False
                        # 2. 语音文件识别, 无法通过openai的audio字段传递，暂时不支持
                        # https://bailian.console.aliyun.com/?tab=doc#/doc/?type=model&url=2979031
                        elif file_type in [
                            'aac',
                            'amr',
                            'aiff',
                            'flac',
                            'm4a',
                            'mp3',
                            'mpeg',
                            'ogg',
                            'opus',
                            'wav',
                            'webm',
                            'wma',
                        ]:
                            me['audio'] = me['file_url']
                            me['type'] = 'audio'
                            del me['file_url']
                            del me['type']
                            del me['file_name']
                            is_use_dashscope_call = True
                            use_time_num += 1
                            use_time_ids.append(message_id)
                            is_enable_multi_model = False
            message_id += 1
        # 使用列表推导式，保留不在 use_time_ids[:-1] 中的元素，仅保留最后一个多媒体消息
        if not is_enable_multi_model and use_time_num > 1:
            messages = [msg for idx, msg in enumerate(messages) if idx not in use_time_ids[:-1]]
        if not is_enable_multi_model:
            messages = [msg for msg in messages if 'resp_message_id' not in msg]
        args['messages'] = messages
        args['stream'] = True
        # 流式处理状态
        # tool_calls_map: dict[str, provider_message.ToolCall] = {}
        chunk_idx = 0
        thinking_started = False
        thinking_ended = False
        role = 'assistant'  # 默认角色
        if is_use_dashscope_call:
            response = dashscope.MultiModalConversation.call(
                # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = "sk-xxx"
                api_key=use_model.provider.token_mgr.get_token(),
                model=use_model.model_entity.name,
                messages=messages,
                result_format='message',
                asr_options={
                    # "language": "zh", # 可选，若已知音频的语种，可通过该参数指定待识别语种，以提升识别准确率
                    'enable_lid': True,
                    'enable_itn': False,
                },
                stream=True,
            )
            content_length_list = []
            previous_length = 0  # 记录上一次的内容长度
            for res in response:
                chunk = res['output']
                # 解析 chunk 数据
                if hasattr(chunk, 'choices') and chunk.choices:
                    choice = chunk.choices[0]
                    delta_content = choice['message'].content[0]['text']
                    finish_reason = choice['finish_reason']
                    content_length_list.append(len(delta_content))
                else:
                    delta_content = ''
                    finish_reason = None
                # 跳过空的第一个 chunk（只有 role 没有内容）
                if chunk_idx == 0 and not delta_content:
                    chunk_idx += 1
                    continue
                # 检查 content_length_list 是否有足够的数据
                if len(content_length_list) >= 2:
                    now_content = delta_content[previous_length : content_length_list[-1]]
                    previous_length = content_length_list[-1]  # 更新上一次的长度
                else:
                    now_content = delta_content  # 第一次循环时直接使用 delta_content
                    previous_length = len(delta_content)  # 更新上一次的长度
                # 构建 MessageChunk - 只包含增量内容
                chunk_data = {
                    'role': role,
                    'content': now_content if now_content else None,
                    'is_final': bool(finish_reason) and finish_reason != 'null',
                }
                # 移除 None 值
                chunk_data = {k: v for k, v in chunk_data.items() if v is not None}
                yield provider_message.MessageChunk(**chunk_data)
                chunk_idx += 1
        else:
            async for chunk in self._req_stream(args, extra_body=extra_args):
                # 解析 chunk 数据
                if hasattr(chunk, 'choices') and chunk.choices:
                    choice = chunk.choices[0]
                    delta = choice.delta.model_dump() if hasattr(choice, 'delta') else {}
                    finish_reason = getattr(choice, 'finish_reason', None)
                else:
                    delta = {}
                    finish_reason = None
                # 从第一个 chunk 获取 role，后续使用这个 role
                if 'role' in delta and delta['role']:
                    role = delta['role']
                # 获取增量内容
                delta_content = delta.get('content', '')
                reasoning_content = delta.get('reasoning_content', '')
                # 处理 reasoning_content
                if reasoning_content:
                    # accumulated_reasoning += reasoning_content
                    # 如果设置了 remove_think，跳过 reasoning_content
                    if remove_think:
                        chunk_idx += 1
                        continue
                    # 第一次出现 reasoning_content，添加 <think> 开始标签
                    if not thinking_started:
                        thinking_started = True
                        delta_content = '<think>\n' + reasoning_content
                    else:
                        # 继续输出 reasoning_content
                        delta_content = reasoning_content
                elif thinking_started and not thinking_ended and delta_content:
                    # reasoning_content 结束，normal content 开始，添加 </think> 结束标签
                    thinking_ended = True
                    delta_content = '\n</think>\n' + delta_content
                # 处理工具调用增量
                if delta.get('tool_calls'):
                    for tool_call in delta['tool_calls']:
                        if tool_call['id'] != '':
                            tool_id = tool_call['id']
                        if tool_call['function']['name'] is not None:
                            tool_name = tool_call['function']['name']
                        if tool_call['type'] is None:
                            tool_call['type'] = 'function'
                        tool_call['id'] = tool_id
                        tool_call['function']['name'] = tool_name
                        tool_call['function']['arguments'] = (
                            '' if tool_call['function']['arguments'] is None else tool_call['function']['arguments']
                        )
                # 跳过空的第一个 chunk（只有 role 没有内容）
                if chunk_idx == 0 and not delta_content and not reasoning_content and not delta.get('tool_calls'):
                    chunk_idx += 1
                    continue
                # 构建 MessageChunk - 只包含增量内容
                chunk_data = {
                    'role': role,
                    'content': delta_content if delta_content else None,
                    'tool_calls': delta.get('tool_calls'),
                    'is_final': bool(finish_reason),
                }
                # 移除 None 值
                chunk_data = {k: v for k, v in chunk_data.items() if v is not None}
                yield provider_message.MessageChunk(**chunk_data)
                chunk_idx += 1
                # return
--- a/src/langbot/pkg/provider/modelmgr/requesters/bailianchatcmpl.yaml
+++ b/src/langbot/pkg/provider/modelmgr/requesters/bailianchatcmpl.yaml
@@ -7,7 +7,6 @@ metadata:
    zh_Hans: 阿里云百炼
  icon: bailian.png
 spec:
  litellm_provider: openai
  config:
  - name: base_url
    label:
@@ -25,7 +24,6 @@ spec:
    default: 120
  support_type:
  - llm
  - text-embedding
  - rerank
  provider_category: maas
 execution:
--- a/src/langbot/pkg/provider/modelmgr/requesters/chatcmpl.py
+++ b/src/langbot/pkg/provider/modelmgr/requesters/chatcmpl.py
@@ -0,0 +1,702 @@
 from __future__ import annotations
 import asyncio
 import typing
 import openai
 import openai.types.chat.chat_completion as chat_completion_module
 import httpx
 from .. import errors, requester
 import langbot_plugin.api.entities.builtin.resource.tool as resource_tool
 import langbot_plugin.api.entities.builtin.pipeline.query as pipeline_query
 import langbot_plugin.api.entities.builtin.provider.message as provider_message
 class OpenAIChatCompletions(requester.ProviderAPIRequester):
    """OpenAI ChatCompletion API 请求器"""
    client: openai.AsyncClient
    default_config: dict[str, typing.Any] = {
        'base_url': 'https://api.openai.com/v1',
        'timeout': 120,
    }
    async def initialize(self):
        self.client = openai.AsyncClient(
            api_key=self.init_api_key,
            base_url=self.requester_cfg['base_url'].replace(' ', ''),
            timeout=self.requester_cfg['timeout'],
            http_client=httpx.AsyncClient(trust_env=True, timeout=self.requester_cfg['timeout']),
        )
    def _mask_api_key(self, api_key: str | None) -> str:
        if not api_key:
            return ''
        if len(api_key) <= 8:
            return '****'
        return f'{api_key[:4]}...{api_key[-4:]}'
    def _infer_model_type(self, model_id: str) -> str:
        normalized_model_id = (model_id or '').lower()
        embedding_keywords = (
            'embedding',
            'embed',
            'bge-',
            'e5-',
            'm3e',
            'gte-',
            'multilingual-e5',
            'text-embedding',
        )
        return 'embedding' if any(keyword in normalized_model_id for keyword in embedding_keywords) else 'llm'
    def _infer_model_abilities(self, item: dict[str, typing.Any], model_id: str) -> list[str]:
        normalized_model_id = (model_id or '').lower()
        abilities: set[str] = set()
        def _flatten(value: typing.Any) -> list[str]:
            if value is None:
                return []
            if isinstance(value, str):
                return [value.lower()]
            if isinstance(value, dict):
                flattened: list[str] = []
                for nested_value in value.values():
                    flattened.extend(_flatten(nested_value))
                return flattened
            if isinstance(value, (list, tuple, set)):
                flattened: list[str] = []
                for nested_value in value:
                    flattened.extend(_flatten(nested_value))
                return flattened
            return [str(value).lower()]
        capability_tokens = _flatten(item.get('capabilities'))
        capability_tokens.extend(_flatten(item.get('modalities')))
        capability_tokens.extend(_flatten(item.get('input_modalities')))
        capability_tokens.extend(_flatten(item.get('output_modalities')))
        capability_tokens.extend(_flatten(item.get('supported_generation_methods')))
        capability_tokens.extend(_flatten(item.get('supported_parameters')))
        capability_tokens.extend(_flatten(item.get('architecture')))
        combined_tokens = capability_tokens + [normalized_model_id]
        vision_keywords = (
            'vision',
            'image',
            'file',
            'video',
            'multimodal',
            'vl',
            'ocr',
            'omni',
        )
        function_call_keywords = (
            'function',
            'tool',
            'tools',
            'tool_choice',
            'tool_call',
            'tool-use',
            'tool_use',
        )
        if any(any(keyword in token for keyword in vision_keywords) for token in combined_tokens):
            abilities.add('vision')
        if any(any(keyword in token for keyword in function_call_keywords) for token in combined_tokens):
            abilities.add('func_call')
        return sorted(abilities)
    def _normalize_modalities(self, value: typing.Any) -> list[str]:
        normalized: list[str] = []
        def _collect(item: typing.Any):
            if item is None:
                return
            if isinstance(item, str):
                for part in item.replace('->', ',').replace('+', ',').split(','):
                    token = part.strip().lower()
                    if token and token not in normalized:
                        normalized.append(token)
                return
            if isinstance(item, dict):
                for nested in item.values():
                    _collect(nested)
                return
            if isinstance(item, (list, tuple, set)):
                for nested in item:
                    _collect(nested)
                return
        _collect(value)
        return normalized
    def _extract_scan_metadata(self, item: dict[str, typing.Any], model_id: str) -> dict[str, typing.Any]:
        display_name = item.get('name')
        if not isinstance(display_name, str) or not display_name.strip() or display_name == model_id:
            display_name = ''
        description = item.get('description')
        if not isinstance(description, str) or not description.strip():
            description = ''
        context_length = item.get('context_length')
        if context_length is None and isinstance(item.get('top_provider'), dict):
            context_length = item['top_provider'].get('context_length')
        if not isinstance(context_length, int):
            try:
                context_length = int(context_length) if context_length is not None else None
            except (TypeError, ValueError):
                context_length = None
        input_modalities = self._normalize_modalities(item.get('input_modalities'))
        output_modalities = self._normalize_modalities(item.get('output_modalities'))
        if isinstance(item.get('architecture'), dict):
            if not input_modalities:
                input_modalities = self._normalize_modalities(item['architecture'].get('input_modalities'))
            if not output_modalities:
                output_modalities = self._normalize_modalities(item['architecture'].get('output_modalities'))
        owned_by = item.get('owned_by')
        if not isinstance(owned_by, str) or not owned_by.strip():
            owned_by = ''
        return {
            'display_name': display_name or None,
            'description': description or None,
            'context_length': context_length,
            'owned_by': owned_by or None,
            'input_modalities': input_modalities,
            'output_modalities': output_modalities,
        }
    async def scan_models(self, api_key: str | None = None) -> dict[str, typing.Any]:
        headers = {}
        if api_key:
            headers['Authorization'] = f'Bearer {api_key}'
        models_url = f'{self.requester_cfg["base_url"].rstrip("/")}/models'
        async with httpx.AsyncClient(trust_env=True, timeout=self.requester_cfg['timeout']) as client:
            response = await client.get(models_url, headers=headers)
            response.raise_for_status()
            payload = response.json()
        models = []
        for item in payload.get('data', []):
            model_id = item.get('id')
            if not model_id:
                continue
            models.append(
                {
                    'id': model_id,
                    'name': model_id,
                    'type': self._infer_model_type(model_id),
                    'abilities': self._infer_model_abilities(item, model_id),
                    **self._extract_scan_metadata(item, model_id),
                }
            )
        models.sort(key=lambda item: (item['type'] != 'llm', item['name'].lower()))
        return {
            'models': models,
            'debug': {
                'request': {
                    'method': 'GET',
                    'url': models_url,
                    'headers': {
                        'Authorization': f'Bearer {self._mask_api_key(api_key)}' if api_key else '',
                    },
                },
                'response': payload,
            },
        }
    async def _req(
        self,
        args: dict,
        extra_body: dict = {},
    ) -> chat_completion_module.ChatCompletion:
        return await self.client.chat.completions.create(**args, extra_body=extra_body)
    async def _req_stream(
        self,
        args: dict,
        extra_body: dict = {},
    ):
        async for chunk in await self.client.chat.completions.create(**args, extra_body=extra_body):
            yield chunk
    async def _make_msg(
        self,
        chat_completion: chat_completion_module.ChatCompletion,
        remove_think: bool = False,
    ) -> provider_message.Message:
        if not isinstance(chat_completion, chat_completion_module.ChatCompletion):
            raise TypeError(f'Expected ChatCompletion, got {type(chat_completion).__name__}: {chat_completion[:16]}')
        chatcmpl_message = chat_completion.choices[0].message.model_dump()
        # 确保 role 字段存在且不为 None
        if 'role' not in chatcmpl_message or chatcmpl_message['role'] is None:
            chatcmpl_message['role'] = 'assistant'
        # 处理思维链
        content = chatcmpl_message.get('content', '')
        reasoning_content = chatcmpl_message.get('reasoning_content', None)
        processed_content, _ = await self._process_thinking_content(
            content=content, reasoning_content=reasoning_content, remove_think=remove_think
        )
        chatcmpl_message['content'] = processed_content
        # 移除 reasoning_content 字段，避免传递给 Message
        if 'reasoning_content' in chatcmpl_message:
            del chatcmpl_message['reasoning_content']
        message = provider_message.Message(**chatcmpl_message)
        return message
    async def _process_thinking_content(
        self,
        content: str,
        reasoning_content: str = None,
        remove_think: bool = False,
    ) -> tuple[str, str]:
        """处理思维链内容
        Args:
            content: 原始内容
            reasoning_content: reasoning_content 字段内容
            remove_think: 是否移除思维链
        Returns:
            (处理后的内容, 提取的思维链内容)
        """
        thinking_content = ''
        # 1. 从 reasoning_content 提取思维链
        if reasoning_content:
            thinking_content = reasoning_content
        # 2. 从 content 中提取 <think> 标签内容
        if content and '<think>' in content and '</think>' in content:
            import re
            think_pattern = r'<think>(.*?)</think>'
            think_matches = re.findall(think_pattern, content, re.DOTALL)
            if think_matches:
                # 如果已有 reasoning_content，则追加
                if thinking_content:
                    thinking_content += '\n' + '\n'.join(think_matches)
                else:
                    thinking_content = '\n'.join(think_matches)
                # 移除 content 中的 <think> 标签
                content = re.sub(think_pattern, '', content, flags=re.DOTALL).strip()
        # 3. 根据 remove_think 参数决定是否保留思维链
        if remove_think:
            return content, ''
        else:
            # 如果有思维链内容，将其以 <think> 格式添加到 content 开头
            if thinking_content:
                content = f'<think>\n{thinking_content}\n</think>\n{content}'.strip()
            return content, thinking_content
    async def _closure_stream(
        self,
        query: pipeline_query.Query,
        req_messages: list[dict],
        use_model: requester.RuntimeLLMModel,
        use_funcs: list[resource_tool.LLMTool] = None,
        extra_args: dict[str, typing.Any] = {},
        remove_think: bool = False,
    ) -> provider_message.MessageChunk:
        self.client.api_key = use_model.provider.token_mgr.get_token()
        args = {}
        args['model'] = use_model.model_entity.name
        if use_funcs:
            tools = await self.ap.tool_mgr.generate_tools_for_openai(use_funcs)
            if tools:
                args['tools'] = tools
        # 设置此次请求中的messages
        messages = req_messages.copy()
        # 检查vision
        for msg in messages:
            if 'content' in msg and isinstance(msg['content'], list):
                for me in msg['content']:
                    if me['type'] == 'image_base64':
                        me['image_url'] = {'url': me['image_base64']}
                        me['type'] = 'image_url'
                        del me['image_base64']
        args['messages'] = messages
        args['stream'] = True
        # 流式处理状态
        # tool_calls_map: dict[str, provider_message.ToolCall] = {}
        chunk_idx = 0
        thinking_started = False
        thinking_ended = False
        role = 'assistant'  # 默认角色
        tool_id = ''
        tool_name = ''
        # accumulated_reasoning = ''  # 仅用于判断何时结束思维链
        async for chunk in self._req_stream(args, extra_body=extra_args):
            # 解析 chunk 数据
            if hasattr(chunk, 'choices') and chunk.choices:
                choice = chunk.choices[0]
                delta = choice.delta.model_dump() if hasattr(choice, 'delta') else {}
                finish_reason = getattr(choice, 'finish_reason', None)
            else:
                delta = {}
                finish_reason = None
            # 从第一个 chunk 获取 role，后续使用这个 role
            if 'role' in delta and delta['role']:
                role = delta['role']
            # 获取增量内容
            delta_content = delta.get('content', '')
            reasoning_content = delta.get('reasoning_content', '')
            # 处理 reasoning_content
            if reasoning_content:
                # accumulated_reasoning += reasoning_content
                # 如果设置了 remove_think，跳过 reasoning_content
                if remove_think:
                    chunk_idx += 1
                    continue
                # 第一次出现 reasoning_content，添加 <think> 开始标签
                if not thinking_started:
                    thinking_started = True
                    delta_content = '<think>\n' + reasoning_content
                else:
                    # 继续输出 reasoning_content
                    delta_content = reasoning_content
            elif thinking_started and not thinking_ended and delta_content:
                # reasoning_content 结束，normal content 开始，添加 </think> 结束标签
                thinking_ended = True
                delta_content = '\n</think>\n' + delta_content
            # 处理 content 中已有的 <think> 标签（如果需要移除）
            # if delta_content and remove_think and '<think>' in delta_content:
            #     import re
            #
            #     # 移除 <think> 标签及其内容
            #     delta_content = re.sub(r'<think>.*?</think>', '', delta_content, flags=re.DOTALL)
            # 处理工具调用增量
            # delta_tool_calls = None
            if delta.get('tool_calls'):
                for tool_call in delta['tool_calls']:
                    if tool_call['id'] and tool_call['function']['name']:
                        tool_id = tool_call['id']
                        tool_name = tool_call['function']['name']
                    else:
                        tool_call['id'] = tool_id
                        tool_call['function']['name'] = tool_name
                    if tool_call['type'] is None:
                        tool_call['type'] = 'function'
            # 跳过空的第一个 chunk（只有 role 没有内容）
            if chunk_idx == 0 and not delta_content and not reasoning_content and not delta.get('tool_calls'):
                chunk_idx += 1
                continue
            # 构建 MessageChunk - 只包含增量内容
            chunk_data = {
                'role': role,
                'content': delta_content if delta_content else None,
                'tool_calls': delta.get('tool_calls'),
                'is_final': bool(finish_reason),
            }
            # 移除 None 值
            chunk_data = {k: v for k, v in chunk_data.items() if v is not None}
            yield provider_message.MessageChunk(**chunk_data)
            chunk_idx += 1
    async def _closure(
        self,
        query: pipeline_query.Query,
        req_messages: list[dict],
        use_model: requester.RuntimeLLMModel,
        use_funcs: list[resource_tool.LLMTool] = None,
        extra_args: dict[str, typing.Any] = {},
        remove_think: bool = False,
    ) -> tuple[provider_message.Message, dict]:
        self.client.api_key = use_model.provider.token_mgr.get_token()
        args = {}
        args['model'] = use_model.model_entity.name
        if use_funcs:
            tools = await self.ap.tool_mgr.generate_tools_for_openai(use_funcs)
            if tools:
                args['tools'] = tools
        # 设置此次请求中的messages
        messages = req_messages.copy()
        # 检查vision
        for msg in messages:
            if 'content' in msg and isinstance(msg['content'], list):
                for me in msg['content']:
                    if me['type'] == 'image_base64':
                        me['image_url'] = {'url': me['image_base64']}
                        me['type'] = 'image_url'
                        del me['image_base64']
        args['messages'] = messages
        # 发送请求
        resp = await self._req(args, extra_body=extra_args)
        # 处理请求结果
        message = await self._make_msg(resp, remove_think)
        # Extract token usage from response
        usage_info = {}
        if hasattr(resp, 'usage') and resp.usage:
            usage_info['input_tokens'] = resp.usage.prompt_tokens or 0
            usage_info['output_tokens'] = resp.usage.completion_tokens or 0
            usage_info['total_tokens'] = resp.usage.total_tokens or 0
        return message, usage_info
    async def invoke_llm(
        self,
        query: pipeline_query.Query,
        model: requester.RuntimeLLMModel,
        messages: typing.List[provider_message.Message],
        funcs: typing.List[resource_tool.LLMTool] = None,
        extra_args: dict[str, typing.Any] = {},
        remove_think: bool = False,
    ) -> tuple[provider_message.Message, dict]:
        """Invoke LLM and return message with usage info"""
        req_messages = []  # req_messages 仅用于类内，外部同步由 query.messages 进行
        for m in messages:
            msg_dict = m.dict(exclude_none=True)
            content = msg_dict.get('content')
            if isinstance(content, list):
                # 检查 content 列表中是否每个部分都是文本
                if all(isinstance(part, dict) and part.get('type') == 'text' for part in content):
                    # 将所有文本部分合并为一个字符串
                    msg_dict['content'] = '\n'.join(part['text'] for part in content)
            req_messages.append(msg_dict)
        try:
            msg, usage_info = await self._closure(
                query=query,
                req_messages=req_messages,
                use_model=model,
                use_funcs=funcs,
                extra_args=extra_args,
                remove_think=remove_think,
            )
            return msg, usage_info
        except asyncio.TimeoutError:
            raise errors.RequesterError('请求超时')
        except openai.BadRequestError as e:
            error_message = str(e.message) if hasattr(e, 'message') else str(e)
            if 'context_length_exceeded' in str(e):
                raise errors.RequesterError(f'上文过长，请重置会话: {error_message}')
            else:
                raise errors.RequesterError(f'请求参数错误: {error_message}')
        except openai.AuthenticationError as e:
            error_message = str(e.message) if hasattr(e, 'message') else str(e)
            raise errors.RequesterError(f'无效的 api-key: {error_message}')
        except openai.NotFoundError as e:
            error_message = str(e.message) if hasattr(e, 'message') else str(e)
            raise errors.RequesterError(f'请求路径错误: {error_message}')
        except openai.RateLimitError as e:
            error_message = str(e.message) if hasattr(e, 'message') else str(e)
            raise errors.RequesterError(f'请求过于频繁或余额不足: {error_message}')
        except openai.APIConnectionError as e:
            error_message = f'连接错误: {str(e)}'
            raise errors.RequesterError(error_message)
        except openai.APIError as e:
            error_message = str(e.message) if hasattr(e, 'message') else str(e)
            raise errors.RequesterError(f'请求错误: {error_message}')
    async def invoke_embedding(
        self,
        model: requester.RuntimeEmbeddingModel,
        input_text: list[str],
        extra_args: dict[str, typing.Any] = {},
    ) -> tuple[list[list[float]], dict]:
        """调用 Embedding API, returns (embeddings, usage_info)"""
        self.client.api_key = model.provider.token_mgr.get_token()
        args = {
            'model': model.model_entity.name,
            'input': input_text,
        }
        if model.model_entity.extra_args:
            args.update(model.model_entity.extra_args)
        args.update(extra_args)
        try:
            resp = await self.client.embeddings.create(**args)
            # Extract usage info
            usage_info = {}
            if hasattr(resp, 'usage') and resp.usage:
                usage_info['prompt_tokens'] = resp.usage.prompt_tokens or 0
                usage_info['total_tokens'] = resp.usage.total_tokens or 0
            return [d.embedding for d in resp.data], usage_info
        except asyncio.TimeoutError:
            raise errors.RequesterError('请求超时')
        except openai.BadRequestError as e:
            raise errors.RequesterError(f'请求参数错误: {e.message}')
    async def invoke_llm_stream(
        self,
        query: pipeline_query.Query,
        model: requester.RuntimeLLMModel,
        messages: typing.List[provider_message.Message],
        funcs: typing.List[resource_tool.LLMTool] = None,
        extra_args: dict[str, typing.Any] = {},
        remove_think: bool = False,
    ) -> provider_message.MessageChunk:
        req_messages = []  # req_messages 仅用于类内，外部同步由 query.messages 进行
        for m in messages:
            msg_dict = m.dict(exclude_none=True)
            content = msg_dict.get('content')
            if isinstance(content, list):
                # 检查 content 列表中是否每个部分都是文本
                if all(isinstance(part, dict) and part.get('type') == 'text' for part in content):
                    # 将所有文本部分合并为一个字符串
                    msg_dict['content'] = '\n'.join(part['text'] for part in content)
            req_messages.append(msg_dict)
        try:
            async for item in self._closure_stream(
                query=query,
                req_messages=req_messages,
                use_model=model,
                use_funcs=funcs,
                extra_args=extra_args,
                remove_think=remove_think,
            ):
                yield item
        except asyncio.TimeoutError:
            raise errors.RequesterError('请求超时')
        except openai.BadRequestError as e:
            if 'context_length_exceeded' in e.message:
                raise errors.RequesterError(f'上文过长，请重置会话: {e.message}')
            else:
                raise errors.RequesterError(f'请求参数错误: {e.message}')
        except openai.AuthenticationError as e:
            raise errors.RequesterError(f'无效的 api-key: {e.message}')
        except openai.NotFoundError as e:
            raise errors.RequesterError(f'请求路径错误: {e.message}')
        except openai.RateLimitError as e:
            raise errors.RequesterError(f'请求过于频繁或余额不足: {e.message}')
        except openai.APIError as e:
            raise errors.RequesterError(f'请求错误: {e.message}')
    async def invoke_rerank(
        self,
        model: requester.RuntimeRerankModel,
        query: str,
        documents: typing.List[str],
        extra_args: dict[str, typing.Any] = {},
    ) -> typing.List[dict]:
        """Standard /rerank endpoint (Jina/Cohere/SiliconFlow/Voyage/DashScope compatible)
        Supports extra_args from model.extra_args:
        - rerank_url: full URL override (e.g. "https://dashscope.aliyuncs.com/compatible-api/v1/reranks")
        - rerank_path: path override appended to base_url (e.g. "reranks" instead of default "rerank")
        - Any other fields are merged into the request payload.
        """
        api_key = model.provider.token_mgr.get_token()
        base_url = self.requester_cfg.get('base_url', '').rstrip('/')
        timeout = self.requester_cfg.get('timeout', 120)
        merged_args = {}
        if model.model_entity.extra_args:
            merged_args.update(model.model_entity.extra_args)
        if extra_args:
            merged_args.update(extra_args)
        rerank_url = merged_args.pop('rerank_url', None)
        rerank_path = merged_args.pop('rerank_path', 'rerank')
        if not rerank_url:
            rerank_url = f'{base_url}/{rerank_path}'
        headers = {
            'Content-Type': 'application/json',
            'Authorization': f'Bearer {api_key}',
        }
        payload = {
            'model': model.model_entity.name,
            'query': query,
            'documents': documents[:64],
            'top_n': min(len(documents), 64),
        }
        if merged_args:
            payload.update(merged_args)
        try:
            async with httpx.AsyncClient(trust_env=True, timeout=timeout) as client:
                resp = await client.post(rerank_url, headers=headers, json=payload)
                resp.raise_for_status()
                data = resp.json()
            results = self._parse_rerank_response(data)
            if results:
                scores = [r.get('relevance_score', 0.0) for r in results]
                min_score = min(scores)
                max_score = max(scores)
                if max_score - min_score > 1e-6:
                    for r in results:
                        r['relevance_score'] = (r['relevance_score'] - min_score) / (max_score - min_score)
            return results
        except httpx.HTTPStatusError as e:
            raise errors.RequesterError(f'Rerank request failed: {e.response.status_code} - {e.response.text}')
        except httpx.TimeoutException:
            raise errors.RequesterError('Rerank request timed out')
        except Exception as e:
            raise errors.RequesterError(f'Rerank request error: {str(e)}')
    @staticmethod
    def _parse_rerank_response(data: dict) -> typing.List[dict]:
        """Parse rerank response from various providers.
        Handles:
        - Jina/Cohere/SiliconFlow: {"results": [{"index", "relevance_score"}]}
        - Voyage AI: {"data": [{"index", "relevance_score"}]}
        - DashScope: {"output": {"results": [{"index", "relevance_score"}]}}
        """
        if 'results' in data:
            return data['results']
        if 'data' in data:
            return data['data']
        if 'output' in data and isinstance(data['output'], dict):
            return data['output'].get('results', [])
        return []
--- a/src/langbot/pkg/provider/modelmgr/requesters/chatcmpl.yaml
+++ b/src/langbot/pkg/provider/modelmgr/requesters/chatcmpl.yaml
@@ -7,7 +7,6 @@ metadata:
    zh_Hans: OpenAI
  icon: openai.svg
 spec:
  litellm_provider: openai
  config:
  - name: base_url
    label:
--- a/src/langbot/pkg/provider/modelmgr/requesters/coherererank.yaml
+++ b/src/langbot/pkg/provider/modelmgr/requesters/coherererank.yaml
@@ -7,7 +7,6 @@ metadata:
    zh_Hans: Cohere
  icon: cohere.svg
 spec:
  litellm_provider: cohere
  config:
  - name: base_url
    label:
--- a/src/langbot/pkg/provider/modelmgr/requesters/compsharechatcmpl.py
+++ b/src/langbot/pkg/provider/modelmgr/requesters/compsharechatcmpl.py
@@ -0,0 +1,17 @@
 from __future__ import annotations
 import typing
 import openai
 from . import chatcmpl
 class CompShareChatCompletions(chatcmpl.OpenAIChatCompletions):
    """CompShare ChatCompletion API 请求器"""
    client: openai.AsyncClient
    default_config: dict[str, typing.Any] = {
        'base_url': 'https://api.modelverse.cn/v1',
        'timeout': 120,
    }
--- a/src/langbot/pkg/provider/modelmgr/requesters/compsharechatcmpl.yaml
+++ b/src/langbot/pkg/provider/modelmgr/requesters/compsharechatcmpl.yaml
@@ -7,7 +7,6 @@ metadata:
    zh_Hans: 优云智算
  icon: compshare.png
 spec:
  litellm_provider: openai
  config:
  - name: base_url
    label:
@@ -25,8 +24,6 @@ spec:
    default: 120
  support_type:
  - llm
  - text-embedding
  - rerank
  provider_category: maas
 execution:
  python:
--- a/src/langbot/pkg/provider/modelmgr/requesters/deepseekchatcmpl.py
+++ b/src/langbot/pkg/provider/modelmgr/requesters/deepseekchatcmpl.py
@@ -0,0 +1,67 @@
 from __future__ import annotations
 import typing
 from . import chatcmpl
 from .. import errors, requester
 import langbot_plugin.api.entities.builtin.resource.tool as resource_tool
 import langbot_plugin.api.entities.builtin.pipeline.query as pipeline_query
 import langbot_plugin.api.entities.builtin.provider.message as provider_message
 class DeepseekChatCompletions(chatcmpl.OpenAIChatCompletions):
    """Deepseek ChatCompletion API 请求器"""
    default_config: dict[str, typing.Any] = {
        'base_url': 'https://api.deepseek.com',
        'timeout': 120,
    }
    async def _closure(
        self,
        query: pipeline_query.Query,
        req_messages: list[dict],
        use_model: requester.RuntimeLLMModel,
        use_funcs: list[resource_tool.LLMTool] = None,
        extra_args: dict[str, typing.Any] = {},
        remove_think: bool = False,
    ) -> tuple[provider_message.Message, dict]:
        self.client.api_key = use_model.provider.token_mgr.get_token()
        args = {}
        args['model'] = use_model.model_entity.name
        if use_funcs:
            tools = await self.ap.tool_mgr.generate_tools_for_openai(use_funcs)
            if tools:
                args['tools'] = tools
        # 设置此次请求中的messages
        messages = req_messages
        # deepseek 不支持多模态，把content都转换成纯文字
        for m in messages:
            if 'content' in m and isinstance(m['content'], list):
                m['content'] = ' '.join([c['text'] for c in m['content'] if 'text' in c])
        args['messages'] = messages
        # 发送请求
        resp = await self._req(args, extra_body=extra_args)
        # print(resp)
        if resp is None:
            raise errors.RequesterError('接口返回为空，请确定模型提供商服务是否正常')
        # 处理请求结果
        message = await self._make_msg(resp, remove_think)
        # Extract token usage from response
        usage_info = {}
        if hasattr(resp, 'usage') and resp.usage:
            usage_info['input_tokens'] = resp.usage.prompt_tokens or 0
            usage_info['output_tokens'] = resp.usage.completion_tokens or 0
            usage_info['total_tokens'] = resp.usage.total_tokens or 0
        return message, usage_info
--- a/src/langbot/pkg/provider/modelmgr/requesters/deepseekchatcmpl.yaml
+++ b/src/langbot/pkg/provider/modelmgr/requesters/deepseekchatcmpl.yaml
@@ -7,7 +7,6 @@ metadata:
    zh_Hans: DeepSeek
  icon: deepseek.svg
 spec:
  litellm_provider: deepseek
  config:
  - name: base_url
    label:
@@ -25,8 +24,6 @@ spec:
    default: 120
  support_type:
  - llm
  - text-embedding
  - rerank
  provider_category: manufacturer
 execution:
  python:
--- a/src/langbot/pkg/provider/modelmgr/requesters/doubao.svg
+++ b/src/langbot/pkg/provider/modelmgr/requesters/doubao.svg
@@ -1,4 +0,0 @@
 <svg width="60" height="50" viewBox="0 0 60 50" xmlns="http://www.w3.org/2000/svg">
  <rect width="60" height="50" rx="8" fill="#3B82F6"/>
  <text x="30" y="32" font-family="Arial, sans-serif" font-size="12" font-weight="bold" fill="white" text-anchor="middle">豆包</text>
 </svg>
--- a/src/langbot/pkg/provider/modelmgr/requesters/doubaochatcmpl.yaml
+++ b/src/langbot/pkg/provider/modelmgr/requesters/doubaochatcmpl.yaml
@@ -1,30 +0,0 @@
 apiVersion: v1
 kind: LLMAPIRequester
 metadata:
  name: doubao-chat-completions
  label:
    en_US: ByteDance Doubao
    zh_Hans: 字节豆包
  icon: doubao.svg
 spec:
  litellm_provider: openai
  config:
  - name: base_url
    label:
      en_US: Base URL
      zh_Hans: 基础 URL
    type: string
    required: true
    default: https://ark.cn-beijing.volces.com/api/v3
  - name: timeout
    label:
      en_US: Timeout
      zh_Hans: 超时时间
    type: integer
    required: true
    default: 120
  support_type:
  - llm
  - text-embedding
  - rerank
  provider_category: manufacturer
--- a/src/langbot/pkg/provider/modelmgr/requesters/geminichatcmpl.py
+++ b/src/langbot/pkg/provider/modelmgr/requesters/geminichatcmpl.py
@@ -0,0 +1,205 @@
 from __future__ import annotations
 import typing
 import httpx
 from . import chatcmpl
 import uuid
 from .. import requester
 import langbot_plugin.api.entities.builtin.provider.message as provider_message
 import langbot_plugin.api.entities.builtin.pipeline.query as pipeline_query
 import langbot_plugin.api.entities.builtin.resource.tool as resource_tool
 class GeminiChatCompletions(chatcmpl.OpenAIChatCompletions):
    """Google Gemini API 请求器"""
    default_config: dict[str, typing.Any] = {
        'base_url': 'https://generativelanguage.googleapis.com/v1beta/openai',
        'timeout': 120,
    }
    async def scan_models(self, api_key: str | None = None) -> dict[str, typing.Any]:
        models_url = 'https://generativelanguage.googleapis.com/v1beta/models'
        params = {'key': api_key} if api_key else {}
        all_models: list[dict[str, typing.Any]] = []
        next_page_token = ''
        last_payload: dict[str, typing.Any] = {}
        async with httpx.AsyncClient(trust_env=True, timeout=self.requester_cfg['timeout']) as client:
            while True:
                request_params = dict(params)
                if next_page_token:
                    request_params['pageToken'] = next_page_token
                response = await client.get(models_url, params=request_params)
                response.raise_for_status()
                payload = response.json()
                last_payload = payload
                for item in payload.get('models', []):
                    model_name = item.get('name', '')
                    model_id = model_name.replace('models/', '', 1)
                    if not model_id:
                        continue
                    supported_methods = item.get('supportedGenerationMethods', []) or []
                    if 'embedContent' in supported_methods and 'generateContent' not in supported_methods:
                        model_type = 'embedding'
                    else:
                        model_type = 'llm'
                    all_models.append(
                        {
                            'id': model_id,
                            'name': model_id,
                            'type': model_type,
                            'abilities': self._infer_model_abilities(item, model_id),
                            'display_name': item.get('displayName') or None,
                            'description': item.get('description') or None,
                            'context_length': item.get('inputTokenLimit'),
                            'input_modalities': self._normalize_modalities(item.get('inputModalities')),
                            'output_modalities': self._normalize_modalities(item.get('outputModalities')),
                        }
                    )
                next_page_token = payload.get('nextPageToken', '')
                if not next_page_token:
                    break
        all_models.sort(key=lambda item: (item['type'] != 'llm', item['name'].lower()))
        return {
            'models': all_models,
            'debug': {
                'request': {
                    'method': 'GET',
                    'url': models_url,
                    'query': {'key': self._mask_api_key(api_key)} if api_key else {},
                },
                'response': last_payload,
            },
        }
    async def _closure_stream(
        self,
        query: pipeline_query.Query,
        req_messages: list[dict],
        use_model: requester.RuntimeLLMModel,
        use_funcs: list[resource_tool.LLMTool] = None,
        extra_args: dict[str, typing.Any] = {},
        remove_think: bool = False,
    ) -> provider_message.MessageChunk:
        self.client.api_key = use_model.provider.token_mgr.get_token()
        args = {}
        args['model'] = use_model.model_entity.name
        if use_funcs:
            tools = await self.ap.tool_mgr.generate_tools_for_openai(use_funcs)
            if tools:
                args['tools'] = tools
        # 设置此次请求中的messages
        messages = req_messages.copy()
        # 检查vision
        for msg in messages:
            if 'content' in msg and isinstance(msg['content'], list):
                for me in msg['content']:
                    if me['type'] == 'image_base64':
                        me['image_url'] = {'url': me['image_base64']}
                        me['type'] = 'image_url'
                        del me['image_base64']
        args['messages'] = messages
        args['stream'] = True
        # 流式处理状态
        # tool_calls_map: dict[str, provider_message.ToolCall] = {}
        chunk_idx = 0
        thinking_started = False
        thinking_ended = False
        role = 'assistant'  # 默认角色
        tool_id = ''
        tool_name = ''
        # accumulated_reasoning = ''  # 仅用于判断何时结束思维链
        async for chunk in self._req_stream(args, extra_body=extra_args):
            # 解析 chunk 数据
            if hasattr(chunk, 'choices') and chunk.choices:
                choice = chunk.choices[0]
                delta = choice.delta.model_dump() if hasattr(choice, 'delta') else {}
                finish_reason = getattr(choice, 'finish_reason', None)
            else:
                delta = {}
                finish_reason = None
            # 从第一个 chunk 获取 role，后续使用这个 role
            if 'role' in delta and delta['role']:
                role = delta['role']
            # 获取增量内容
            delta_content = delta.get('content', '')
            reasoning_content = delta.get('reasoning_content', '')
            # 处理 reasoning_content
            if reasoning_content:
                # accumulated_reasoning += reasoning_content
                # 如果设置了 remove_think，跳过 reasoning_content
                if remove_think:
                    chunk_idx += 1
                    continue
                # 第一次出现 reasoning_content，添加 <think> 开始标签
                if not thinking_started:
                    thinking_started = True
                    delta_content = '<think>\n' + reasoning_content
                else:
                    # 继续输出 reasoning_content
                    delta_content = reasoning_content
            elif thinking_started and not thinking_ended and delta_content:
                # reasoning_content 结束，normal content 开始，添加 </think> 结束标签
                thinking_ended = True
                delta_content = '\n</think>\n' + delta_content
            # 处理 content 中已有的 <think> 标签（如果需要移除）
            # if delta_content and remove_think and '<think>' in delta_content:
            #     import re
            #
            #     # 移除 <think> 标签及其内容
            #     delta_content = re.sub(r'<think>.*?</think>', '', delta_content, flags=re.DOTALL)
            # 处理工具调用增量
            # delta_tool_calls = None
            if delta.get('tool_calls'):
                for tool_call in delta['tool_calls']:
                    if tool_call['id'] == '' and tool_id == '':
                        tool_id = str(uuid.uuid4())
                    if tool_call['function']['name']:
                        tool_name = tool_call['function']['name']
                    tool_call['id'] = tool_id
                    tool_call['function']['name'] = tool_name
                    if tool_call['type'] is None:
                        tool_call['type'] = 'function'
            # 跳过空的第一个 chunk（只有 role 没有内容）
            if chunk_idx == 0 and not delta_content and not reasoning_content and not delta.get('tool_calls'):
                chunk_idx += 1
                continue
            # 构建 MessageChunk - 只包含增量内容
            chunk_data = {
                'role': role,
                'content': delta_content if delta_content else None,
                'tool_calls': delta.get('tool_calls'),
                'is_final': bool(finish_reason),
            }
            # 移除 None 值
            chunk_data = {k: v for k, v in chunk_data.items() if v is not None}
            yield provider_message.MessageChunk(**chunk_data)
            chunk_idx += 1
--- a/src/langbot/pkg/provider/modelmgr/requesters/geminichatcmpl.yaml
+++ b/src/langbot/pkg/provider/modelmgr/requesters/geminichatcmpl.yaml
@@ -7,7 +7,6 @@ metadata:
    zh_Hans: Google Gemini
  icon: gemini.svg
 spec:
  litellm_provider: gemini
  config:
  - name: base_url
    label:
@@ -25,8 +24,6 @@ spec:
    default: 120
  support_type:
  - llm
  - text-embedding
  - rerank
  provider_category: manufacturer
 execution:
  python:
--- a/src/langbot/pkg/provider/modelmgr/requesters/giteeaichatcmpl.py
+++ b/src/langbot/pkg/provider/modelmgr/requesters/giteeaichatcmpl.py
@@ -0,0 +1,15 @@
 from __future__ import annotations
 import typing
 from . import ppiochatcmpl
 class GiteeAIChatCompletions(ppiochatcmpl.PPIOChatCompletions):
    """Gitee AI ChatCompletions API 请求器"""
    default_config: dict[str, typing.Any] = {
        'base_url': 'https://ai.gitee.com/v1',
        'timeout': 120,
    }
--- a/src/langbot/pkg/provider/modelmgr/requesters/giteeaichatcmpl.yaml
+++ b/src/langbot/pkg/provider/modelmgr/requesters/giteeaichatcmpl.yaml
@@ -7,7 +7,6 @@ metadata:
    zh_Hans: Gitee AI
  icon: giteeai.svg
 spec:
  litellm_provider: openai
  config:
  - name: base_url
    label:
--- a/src/langbot/pkg/provider/modelmgr/requesters/groq.svg
+++ b/src/langbot/pkg/provider/modelmgr/requesters/groq.svg
@@ -1,4 +0,0 @@
 <svg width="60" height="50" viewBox="0 0 60 50" xmlns="http://www.w3.org/2000/svg">
  <rect width="60" height="50" rx="8" fill="#F97316"/>
  <text x="30" y="32" font-family="Arial, sans-serif" font-size="14" font-weight="bold" fill="white" text-anchor="middle">Groq</text>
 </svg>
--- a/src/langbot/pkg/provider/modelmgr/requesters/iflytek.svg
+++ b/src/langbot/pkg/provider/modelmgr/requesters/iflytek.svg
@@ -1,5 +0,0 @@
 <svg width="60" height="50" viewBox="0 0 60 50" xmlns="http://www.w3.org/2000/svg">
  <rect width="60" height="50" rx="8" fill="#0066FF"/>
  <text x="30" y="28" font-family="Arial, sans-serif" font-size="10" font-weight="bold" fill="white" text-anchor="middle">iFlytek</text>
  <text x="30" y="40" font-family="Arial, sans-serif" font-size="8" fill="white" text-anchor="middle">Spark</text>
 </svg>
--- a/src/langbot/pkg/provider/modelmgr/requesters/iflytekchatcmpl.yaml
+++ b/src/langbot/pkg/provider/modelmgr/requesters/iflytekchatcmpl.yaml
@@ -1,30 +0,0 @@
 apiVersion: v1
 kind: LLMAPIRequester
 metadata:
  name: iflytek-chat-completions
  label:
    en_US: iFlytek Spark
    zh_Hans: 讯飞星火
  icon: iflytek.svg
 spec:
  litellm_provider: openai
  config:
  - name: base_url
    label:
      en_US: Base URL
      zh_Hans: 基础 URL
    type: string
    required: true
    default: https://spark-api-open.xf-yun.com/v1
  - name: timeout
    label:
      en_US: Timeout
      zh_Hans: 超时时间
    type: integer
    required: true
    default: 120
  support_type:
  - llm
  - text-embedding
  - rerank
  provider_category: manufacturer
--- a/src/langbot/pkg/provider/modelmgr/requesters/jiekouaichatcmpl.py
+++ b/src/langbot/pkg/provider/modelmgr/requesters/jiekouaichatcmpl.py
@@ -0,0 +1,208 @@
 from __future__ import annotations
 import openai
 import typing
 from . import chatcmpl
 from .. import requester
 import openai.types.chat.chat_completion as chat_completion
 import re
 import langbot_plugin.api.entities.builtin.provider.message as provider_message
 import langbot_plugin.api.entities.builtin.pipeline.query as pipeline_query
 import langbot_plugin.api.entities.builtin.resource.tool as resource_tool
 class JieKouAIChatCompletions(chatcmpl.OpenAIChatCompletions):
    """接口 AI ChatCompletion API 请求器"""
    client: openai.AsyncClient
    default_config: dict[str, typing.Any] = {
        'base_url': 'https://api.jiekou.ai/openai',
        'timeout': 120,
    }
    is_think: bool = False
    async def _make_msg(
        self,
        chat_completion: chat_completion.ChatCompletion,
        remove_think: bool,
    ) -> provider_message.Message:
        chatcmpl_message = chat_completion.choices[0].message.model_dump()
        # print(chatcmpl_message.keys(), chatcmpl_message.values())
        # 确保 role 字段存在且不为 None
        if 'role' not in chatcmpl_message or chatcmpl_message['role'] is None:
            chatcmpl_message['role'] = 'assistant'
        reasoning_content = chatcmpl_message['reasoning_content'] if 'reasoning_content' in chatcmpl_message else None
        # deepseek的reasoner模型
        chatcmpl_message['content'] = await self._process_thinking_content(
            chatcmpl_message['content'], reasoning_content, remove_think
        )
        # 移除 reasoning_content 字段，避免传递给 Message
        if 'reasoning_content' in chatcmpl_message:
            del chatcmpl_message['reasoning_content']
        message = provider_message.Message(**chatcmpl_message)
        return message
    async def _process_thinking_content(
        self,
        content: str,
        reasoning_content: str = None,
        remove_think: bool = False,
    ) -> tuple[str, str]:
        """处理思维链内容
        Args:
            content: 原始内容
            reasoning_content: reasoning_content 字段内容
            remove_think: 是否移除思维链
        Returns:
            处理后的内容
        """
        if remove_think:
            content = re.sub(r'<think>.*?</think>', '', content, flags=re.DOTALL)
        else:
            if reasoning_content is not None:
                content = '<think>\n' + reasoning_content + '\n</think>\n' + content
        return content
    async def _make_msg_chunk(
        self,
        delta: dict[str, typing.Any],
        idx: int,
    ) -> provider_message.MessageChunk:
        # 处理流式chunk和完整响应的差异
        # print(chat_completion.choices[0])
        # 确保 role 字段存在且不为 None
        if 'role' not in delta or delta['role'] is None:
            delta['role'] = 'assistant'
        reasoning_content = delta['reasoning_content'] if 'reasoning_content' in delta else None
        delta['content'] = '' if delta['content'] is None else delta['content']
        # print(reasoning_content)
        # deepseek的reasoner模型
        if reasoning_content is not None:
            delta['content'] += reasoning_content
        message = provider_message.MessageChunk(**delta)
        return message
    async def _closure_stream(
        self,
        query: pipeline_query.Query,
        req_messages: list[dict],
        use_model: requester.RuntimeLLMModel,
        use_funcs: list[resource_tool.LLMTool] = None,
        extra_args: dict[str, typing.Any] = {},
        remove_think: bool = False,
    ) -> provider_message.Message | typing.AsyncGenerator[provider_message.MessageChunk, None]:
        self.client.api_key = use_model.provider.token_mgr.get_token()
        args = {}
        args['model'] = use_model.model_entity.name
        if use_funcs:
            tools = await self.ap.tool_mgr.generate_tools_for_openai(use_funcs)
            if tools:
                args['tools'] = tools
        # 设置此次请求中的messages
        messages = req_messages.copy()
        # 检查vision
        for msg in messages:
            if 'content' in msg and isinstance(msg['content'], list):
                for me in msg['content']:
                    if me['type'] == 'image_base64':
                        me['image_url'] = {'url': me['image_base64']}
                        me['type'] = 'image_url'
                        del me['image_base64']
        args['messages'] = messages
        args['stream'] = True
        # tool_calls_map: dict[str, provider_message.ToolCall] = {}
        chunk_idx = 0
        thinking_started = False
        thinking_ended = False
        role = 'assistant'  # 默认角色
        async for chunk in self._req_stream(args, extra_body=extra_args):
            # 解析 chunk 数据
            if hasattr(chunk, 'choices') and chunk.choices:
                choice = chunk.choices[0]
                delta = choice.delta.model_dump() if hasattr(choice, 'delta') else {}
                finish_reason = getattr(choice, 'finish_reason', None)
            else:
                delta = {}
                finish_reason = None
            # 从第一个 chunk 获取 role，后续使用这个 role
            if 'role' in delta and delta['role']:
                role = delta['role']
            # 获取增量内容
            delta_content = delta.get('content', '')
            # reasoning_content = delta.get('reasoning_content', '')
            if remove_think:
                if delta['content'] is not None:
                    if '<think>' in delta['content'] and not thinking_started and not thinking_ended:
                        thinking_started = True
                        continue
                    elif delta['content'] == r'</think>' and not thinking_ended:
                        thinking_ended = True
                        continue
                    elif thinking_ended and delta['content'] == '\n\n' and thinking_started:
                        thinking_started = False
                        continue
                    elif thinking_started and not thinking_ended:
                        continue
            # delta_tool_calls = None
            if delta.get('tool_calls'):
                for tool_call in delta['tool_calls']:
                    if tool_call['id'] and tool_call['function']['name']:
                        tool_id = tool_call['id']
                        tool_name = tool_call['function']['name']
                    if tool_call['id'] is None:
                        tool_call['id'] = tool_id
                    if tool_call['function']['name'] is None:
                        tool_call['function']['name'] = tool_name
                    if tool_call['function']['arguments'] is None:
                        tool_call['function']['arguments'] = ''
                    if tool_call['type'] is None:
                        tool_call['type'] = 'function'
            # 跳过空的第一个 chunk（只有 role 没有内容）
            if chunk_idx == 0 and not delta_content and not delta.get('tool_calls'):
                chunk_idx += 1
                continue
            # 构建 MessageChunk - 只包含增量内容
            chunk_data = {
                'role': role,
                'content': delta_content if delta_content else None,
                'tool_calls': delta.get('tool_calls'),
                'is_final': bool(finish_reason),
            }
            # 移除 None 值
            chunk_data = {k: v for k, v in chunk_data.items() if v is not None}
            yield provider_message.MessageChunk(**chunk_data)
            chunk_idx += 1
--- a/src/langbot/pkg/provider/modelmgr/requesters/jiekouaichatcmpl.yaml
+++ b/src/langbot/pkg/provider/modelmgr/requesters/jiekouaichatcmpl.yaml
@@ -7,7 +7,6 @@ metadata:
    zh_Hans: 接口 AI
  icon: jiekouai.png
 spec:
  litellm_provider: openai
  config:
  - name: base_url
    label:
--- a/src/langbot/pkg/provider/modelmgr/requesters/jinarerank.yaml
+++ b/src/langbot/pkg/provider/modelmgr/requesters/jinarerank.yaml
@@ -7,7 +7,6 @@ metadata:
    zh_Hans: Jina
  icon: jina.svg
 spec:
  litellm_provider: openai
  config:
  - name: base_url
    label:
--- a/src/langbot/pkg/provider/modelmgr/requesters/litellmchat.py
+++ b/src/langbot/pkg/provider/modelmgr/requesters/litellmchat.py
@@ -1,407 +0,0 @@
 """LiteLLM unified requester for chat, embedding, and rerank."""
 from __future__ import annotations
 import typing
 import litellm
 from litellm import acompletion, aembedding, arerank
 from .. import errors, requester
 import langbot_plugin.api.entities.builtin.resource.tool as resource_tool
 import langbot_plugin.api.entities.builtin.pipeline.query as pipeline_query
 import langbot_plugin.api.entities.builtin.provider.message as provider_message
 class LiteLLMRequester(requester.ProviderAPIRequester):
    """LiteLLM unified API requester supporting chat, embedding, and rerank."""
    default_config: dict[str, typing.Any] = {
        'base_url': '',
        'timeout': 120,
        'custom_llm_provider': '',
        'drop_params': False,
        'num_retries': 0,
        'api_version': '',
    }
    async def initialize(self):
        """Initialize LiteLLM client settings."""
        # LiteLLM doesn't require explicit client initialization
        # Configuration is passed per-request via litellm params
        pass
    def _build_litellm_model_name(self, model_name: str, custom_llm_provider: str | None = None) -> str:
        """Build LiteLLM model name with provider prefix if needed."""
        provider = custom_llm_provider or self.requester_cfg.get('custom_llm_provider', '')
        if provider:
            # LiteLLM format: provider/model_name
            return f'{provider}/{model_name}'
        # If no custom provider, assume model_name already includes prefix or is OpenAI-compatible
        return model_name
    def _convert_messages(self, messages: typing.List[provider_message.Message]) -> list[dict]:
        """Convert LangBot messages to LiteLLM/OpenAI format."""
        req_messages = []
        for m in messages:
            msg_dict = m.dict(exclude_none=True)
            content = msg_dict.get('content')
            if isinstance(content, list):
                for part in content:
                    if isinstance(part, dict) and part.get('type') == 'image_base64':
                        part['image_url'] = {'url': part['image_base64']}
                        part['type'] = 'image_url'
                        del part['image_base64']
            req_messages.append(msg_dict)
        return req_messages
    def _process_thinking_content(self, content: str, reasoning_content: str | None, remove_think: bool) -> str:
        """Process thinking/reasoning content.
        Args:
            content: The main content from response
            reasoning_content: Separate reasoning content from model
            remove_think: If True, remove thinking markers; if False, preserve them
        Returns:
            Processed content string
        """
        # Extract and handle thinking tags
        if content and 'CRETIRE_REASONING_BEGINk' in content and 'CRETIRE_REASONING_ENDk' in content:
            import re
            think_pattern = r'CRETIRE_REASONING_BEGINk(.*?)CRETIRE_REASONING_ENDk'
            if remove_think:
                # Remove thinking tags and their content from output
                content = re.sub(think_pattern, '', content, flags=re.DOTALL).strip()
            # else: preserve thinking content as-is
        # Handle separate reasoning_content field
        # Currently we don't include reasoning_content in user-facing output regardless of remove_think
        # because it's typically internal model reasoning, not user-visible thinking
        return content or ''
    def _extract_usage(self, response) -> dict:
        """Extract usage info from LiteLLM response."""
        usage = response.usage
        return {
            'prompt_tokens': usage.prompt_tokens or 0,
            'completion_tokens': usage.completion_tokens or 0,
            'total_tokens': usage.total_tokens or 0,
        }
    def _build_common_args(self, args: dict, include_retry_params: bool = True) -> dict:
        """Apply common requester config to args dict."""
        if self.requester_cfg.get('base_url'):
            args['api_base'] = self.requester_cfg['base_url']
        if self.requester_cfg.get('timeout'):
            args['timeout'] = self.requester_cfg['timeout']
        if include_retry_params:
            if self.requester_cfg.get('drop_params'):
                args['drop_params'] = self.requester_cfg['drop_params']
            if self.requester_cfg.get('num_retries'):
                args['num_retries'] = self.requester_cfg['num_retries']
            if self.requester_cfg.get('api_version'):
                args['api_version'] = self.requester_cfg['api_version']
        return args
    def _handle_litellm_error(self, e: Exception) -> None:
        """Convert LiteLLM exceptions to RequesterError. Never returns, always raises."""
        # Check more specific exceptions first (they inherit from base exceptions)
        if isinstance(e, litellm.ContextWindowExceededError):
            raise errors.RequesterError(f'上下文长度超限: {str(e)}')
        if isinstance(e, litellm.BadRequestError):
            raise errors.RequesterError(f'请求参数错误: {str(e)}')
        if isinstance(e, litellm.AuthenticationError):
            raise errors.RequesterError(f'API key 无效: {str(e)}')
        if isinstance(e, litellm.NotFoundError):
            raise errors.RequesterError(f'模型或路径无效: {str(e)}')
        if isinstance(e, litellm.RateLimitError):
            raise errors.RequesterError(f'请求过于频繁或余额不足: {str(e)}')
        if isinstance(e, litellm.Timeout):
            raise errors.RequesterError(f'请求超时: {str(e)}')
        if isinstance(e, litellm.APIConnectionError):
            raise errors.RequesterError(f'连接错误: {str(e)}')
        if isinstance(e, litellm.APIError):
            raise errors.RequesterError(f'API 错误: {str(e)}')
        raise errors.RequesterError(f'未知错误: {str(e)}')
    async def _build_completion_args(
        self,
        model: requester.RuntimeLLMModel,
        messages: typing.List[provider_message.Message],
        funcs: typing.List[resource_tool.LLMTool] = None,
        extra_args: dict[str, typing.Any] = {},
        stream: bool = False,
    ) -> dict:
        """Build common completion arguments for invoke_llm and invoke_llm_stream."""
        req_messages = self._convert_messages(messages)
        model_name = self._build_litellm_model_name(model.model_entity.name)
        api_key = model.provider.token_mgr.get_token()
        args = {
            'model': model_name,
            'messages': req_messages,
            'api_key': api_key,
        }
        if stream:
            args['stream'] = True
            args['stream_options'] = {'include_usage': True}
        self._build_common_args(args)
        # Apply model-level extra_args first, then call-level extra_args
        if model.model_entity.extra_args:
            args.update(model.model_entity.extra_args)
        args.update(extra_args)
        if funcs:
            tools = await self.ap.tool_mgr.generate_tools_for_openai(funcs)
            if tools:
                args['tools'] = tools
        return args
    async def invoke_llm(
        self,
        query: pipeline_query.Query,
        model: requester.RuntimeLLMModel,
        messages: typing.List[provider_message.Message],
        funcs: typing.List[resource_tool.LLMTool] = None,
        extra_args: dict[str, typing.Any] = {},
        remove_think: bool = False,
    ) -> tuple[provider_message.Message, dict]:
        """Invoke LLM and return message with usage info."""
        args = await self._build_completion_args(model, messages, funcs, extra_args, stream=False)
        try:
            response = await acompletion(**args)
            message_data = response.choices[0].message.model_dump()
            if 'role' not in message_data or message_data['role'] is None:
                message_data['role'] = 'assistant'
            content = message_data.get('content', '')
            reasoning_content = message_data.get('reasoning_content', None)
            message_data['content'] = self._process_thinking_content(content, reasoning_content, remove_think)
            if 'reasoning_content' in message_data:
                del message_data['reasoning_content']
            message = provider_message.Message(**message_data)
            usage_info = self._extract_usage(response)
            return message, usage_info
        except Exception as e:
            self._handle_litellm_error(e)
    async def invoke_llm_stream(
        self,
        query: pipeline_query.Query,
        model: requester.RuntimeLLMModel,
        messages: typing.List[provider_message.Message],
        funcs: typing.List[resource_tool.LLMTool] = None,
        extra_args: dict[str, typing.Any] = {},
        remove_think: bool = False,
    ) -> provider_message.MessageChunk:
        """Invoke LLM streaming and yield chunks."""
        args = await self._build_completion_args(model, messages, funcs, extra_args, stream=True)
        chunk_idx = 0
        role = 'assistant'
        try:
            response = await acompletion(**args)
            async for chunk in response:
                # Check for usage chunk (final chunk with stream_options include_usage)
                if hasattr(chunk, 'usage') and chunk.usage and (not hasattr(chunk, 'choices') or not chunk.choices):
                    usage_info = {
                        'prompt_tokens': chunk.usage.prompt_tokens or 0,
                        'completion_tokens': chunk.usage.completion_tokens or 0,
                        'total_tokens': chunk.usage.total_tokens or 0,
                    }
                    if query:
                        if query.variables is None:
                            query.variables = {}
                        query.variables['_stream_usage'] = usage_info
                    continue
                if not hasattr(chunk, 'choices') or not chunk.choices:
                    continue
                choice = chunk.choices[0]
                delta = choice.delta.model_dump() if hasattr(choice, 'delta') else {}
                finish_reason = getattr(choice, 'finish_reason', None)
                if 'role' in delta and delta['role']:
                    role = delta['role']
                delta_content = delta.get('content', '')
                reasoning_content = delta.get('reasoning_content', '')
                # Handle reasoning_content based on remove_think flag
                if reasoning_content:
                    if remove_think:
                        # Skip reasoning content when remove_think is True
                        chunk_idx += 1
                        continue
                    else:
                        # Use reasoning_content as the displayed content
                        delta_content = reasoning_content
                if chunk_idx == 0 and not delta_content and not delta.get('tool_calls'):
                    chunk_idx += 1
                    continue
                chunk_data = {
                    'role': role,
                    'content': delta_content if delta_content else None,
                    'tool_calls': delta.get('tool_calls'),
                    'is_final': bool(finish_reason),
                }
                chunk_data = {k: v for k, v in chunk_data.items() if v is not None}
                yield provider_message.MessageChunk(**chunk_data)
                chunk_idx += 1
        except Exception as e:
            self._handle_litellm_error(e)
    async def invoke_embedding(
        self,
        model: requester.RuntimeEmbeddingModel,
        input_text: list[str],
        extra_args: dict[str, typing.Any] = {},
    ) -> tuple[list[list[float]], dict]:
        """Invoke embedding and return vectors with usage info."""
        model_name = self._build_litellm_model_name(model.model_entity.name)
        api_key = model.provider.token_mgr.get_token()
        args = {
            'model': model_name,
            'input': input_text,
            'api_key': api_key,
        }
        self._build_common_args(args, include_retry_params=False)
        if model.model_entity.extra_args:
            args.update(model.model_entity.extra_args)
        args.update(extra_args)
        try:
            response = await aembedding(**args)
            embeddings = [d.embedding for d in response.data]
            usage_info = self._extract_usage(response)
            return embeddings, usage_info
        except Exception as e:
            self._handle_litellm_error(e)
    async def invoke_rerank(
        self,
        model: requester.RuntimeRerankModel,
        query: str,
        documents: typing.List[str],
        extra_args: dict[str, typing.Any] = {},
    ) -> typing.List[dict]:
        """Invoke rerank and return relevance scores."""
        model_name = self._build_litellm_model_name(model.model_entity.name)
        api_key = model.provider.token_mgr.get_token()
        args = {
            'model': model_name,
            'query': query,
            'documents': documents,
            'api_key': api_key,
            'top_n': min(len(documents), 64),
        }
        self._build_common_args(args, include_retry_params=False)
        if model.model_entity.extra_args:
            args.update(model.model_entity.extra_args)
        args.update(extra_args)
        try:
            response = await arerank(**args)
            results = []
            for r in response.results:
                results.append(
                    {
                        'index': r.get('index', 0),
                        'relevance_score': r.get('relevance_score', 0.0),
                    }
                )
            if results:
                scores = [r['relevance_score'] for r in results]
                min_score = min(scores)
                max_score = max(scores)
                if max_score - min_score > 1e-6:
                    for r in results:
                        r['relevance_score'] = (r['relevance_score'] - min_score) / (max_score - min_score)
            return results
        except Exception as e:
            self._handle_litellm_error(e)
    async def scan_models(self, api_key: str | None = None) -> dict[str, typing.Any]:
        """Scan models supported by the provider."""
        import httpx
        base_url = self.requester_cfg.get('base_url', '').rstrip('/')
        timeout = self.requester_cfg.get('timeout', 120)
        if not base_url:
            raise errors.RequesterError('Base URL required for model scanning')
        headers = {}
        if api_key:
            headers['Authorization'] = f'Bearer {api_key}'
        models_url = f'{base_url}/models'
        try:
            async with httpx.AsyncClient(trust_env=True, timeout=timeout) as client:
                response = await client.get(models_url, headers=headers)
                response.raise_for_status()
                payload = response.json()
            models = []
            for item in payload.get('data', []):
                model_id = item.get('id')
                if not model_id:
                    continue
                # Infer model type
                normalized_id = (model_id or '').lower()
                embedding_keywords = ('embedding', 'embed', 'bge-', 'e5-', 'm3e', 'gte-', 'text-embedding')
                model_type = 'embedding' if any(kw in normalized_id for kw in embedding_keywords) else 'llm'
                models.append(
                    {
                        'id': model_id,
                        'name': model_id,
                        'type': model_type,
                    }
                )
            models.sort(key=lambda x: (x['type'] != 'llm', x['name'].lower()))
            return {'models': models}
        except httpx.HTTPStatusError as e:
            raise errors.RequesterError(f'Model scan failed: {e.response.status_code}')
        except httpx.TimeoutException:
            raise errors.RequesterError('Model scan timeout')
        except Exception as e:
            raise errors.RequesterError(f'Model scan error: {str(e)}')
--- a/src/langbot/pkg/provider/modelmgr/requesters/litellmchat.yaml
+++ b/src/langbot/pkg/provider/modelmgr/requesters/litellmchat.yaml
@@ -1,64 +0,0 @@
 apiVersion: v1
 kind: LLMAPIRequester
 metadata:
  name: litellm-chat
  label:
    en_US: LiteLLM (Unified)
    zh_Hans: LiteLLM (统一请求器)
  icon: litellm.svg
 spec:
  config:
  - name: base_url
    label:
      en_US: Base URL
      zh_Hans: 基础 URL
    type: string
    required: false
    default: ''
  - name: timeout
    label:
      en_US: Timeout
      zh_Hans: 超时时间
    type: integer
    required: true
    default: 120
  - name: custom_llm_provider
    label:
      en_US: Custom Provider
      zh_Hans: 自定义 Provider
    type: string
    required: false
    default: ''
    description:
      en_US: Force provider type (e.g., anthropic, openai, gemini)
      zh_Hans: 强制指定 provider 类型（如 anthropic, openai, gemini）
  - name: drop_params
    label:
      en_US: Drop Unsupported Params
      zh_Hans: 丢弃不支持参数
    type: boolean
    required: false
    default: false
  - name: num_retries
    label:
      en_US: Number of Retries
      zh_Hans: 重试次数
    type: integer
    required: false
    default: 0
  - name: api_version
    label:
      en_US: API Version
      zh_Hans: API 版本
    type: string
    required: false
    default: ''
  support_type:
  - llm
  - text-embedding
  - rerank
  provider_category: unified
 execution:
  python:
    path: ./litellmchat.py
    attr: LiteLLMRequester
--- a/src/langbot/pkg/provider/modelmgr/requesters/lmstudiochatcmpl.py
+++ b/src/langbot/pkg/provider/modelmgr/requesters/lmstudiochatcmpl.py
@@ -0,0 +1,17 @@
 from __future__ import annotations
 import typing
 import openai
 from . import chatcmpl
 class LmStudioChatCompletions(chatcmpl.OpenAIChatCompletions):
    """LMStudio ChatCompletion API 请求器"""
    client: openai.AsyncClient
    default_config: dict[str, typing.Any] = {
        'base_url': 'http://127.0.0.1:1234/v1',
        'timeout': 120,
    }
--- a/src/langbot/pkg/provider/modelmgr/requesters/lmstudiochatcmpl.yaml
+++ b/src/langbot/pkg/provider/modelmgr/requesters/lmstudiochatcmpl.yaml
@@ -7,7 +7,6 @@ metadata:
    zh_Hans: LM Studio
  icon: lmstudio.webp
 spec:
  litellm_provider: openai
  config:
  - name: base_url
    label:
--- a/src/langbot/pkg/provider/modelmgr/requesters/mimo.svg
+++ b/src/langbot/pkg/provider/modelmgr/requesters/mimo.svg
@@ -1,4 +0,0 @@
 <svg width="60" height="50" viewBox="0 0 60 50" xmlns="http://www.w3.org/2000/svg">
  <rect width="60" height="50" rx="8" fill="#FF6700"/>
  <text x="30" y="32" font-family="Arial, sans-serif" font-size="18" font-weight="bold" fill="white" text-anchor="middle">MiMo</text>
 </svg>
--- a/src/langbot/pkg/provider/modelmgr/requesters/mimochatcmpl.yaml
+++ b/src/langbot/pkg/provider/modelmgr/requesters/mimochatcmpl.yaml
@@ -1,30 +0,0 @@
 apiVersion: v1
 kind: LLMAPIRequester
 metadata:
  name: mimo-chat-completions
  label:
    en_US: Xiaomi MiMo
    zh_Hans: 小米 MiMo
  icon: mimo.svg
 spec:
  litellm_provider: openai
  config:
  - name: base_url
    label:
      en_US: Base URL
      zh_Hans: 基础 URL
    type: string
    required: true
    default: https://api.xiaomimimo.com/v1
  - name: timeout
    label:
      en_US: Timeout
      zh_Hans: 超时时间
    type: integer
    required: true
    default: 120
  support_type:
  - llm
  - text-embedding
  - rerank
  provider_category: manufacturer
--- a/src/langbot/pkg/provider/modelmgr/requesters/minimax.svg
+++ b/src/langbot/pkg/provider/modelmgr/requesters/minimax.svg
@@ -1,4 +0,0 @@
 <svg width="60" height="50" viewBox="0 0 60 50" xmlns="http://www.w3.org/2000/svg">
  <rect width="60" height="50" rx="8" fill="#4F46E5"/>
  <text x="30" y="32" font-family="Arial, sans-serif" font-size="12" font-weight="bold" fill="white" text-anchor="middle">MiniMax</text>
 </svg>
--- a/src/langbot/pkg/provider/modelmgr/requesters/minimaxchatcmpl.yaml
+++ b/src/langbot/pkg/provider/modelmgr/requesters/minimaxchatcmpl.yaml
@@ -1,30 +0,0 @@
 apiVersion: v1
 kind: LLMAPIRequester
 metadata:
  name: minimax-chat-completions
  label:
    en_US: MiniMax
    zh_Hans: MiniMax
  icon: minimax.svg
 spec:
  litellm_provider: openai
  config:
  - name: base_url
    label:
      en_US: Base URL
      zh_Hans: 基础 URL
    type: string
    required: true
    default: https://api.minimax.chat/v1
  - name: timeout
    label:
      en_US: Timeout
      zh_Hans: 超时时间
    type: integer
    required: true
    default: 120
  support_type:
  - llm
  - text-embedding
  - rerank
  provider_category: manufacturer
--- a/src/langbot/pkg/provider/modelmgr/requesters/mistral.svg
+++ b/src/langbot/pkg/provider/modelmgr/requesters/mistral.svg
@@ -1,5 +0,0 @@
 <svg width="60" height="50" viewBox="0 0 60 50" xmlns="http://www.w3.org/2000/svg">
  <rect width="60" height="50" rx="8" fill="#FF6B35"/>
  <text x="30" y="28" font-family="Arial, sans-serif" font-size="10" font-weight="bold" fill="white" text-anchor="middle">Mistral</text>
  <text x="30" y="40" font-family="Arial, sans-serif" font-size="8" fill="white" text-anchor="middle">AI</text>
 </svg>
--- a/src/langbot/pkg/provider/modelmgr/requesters/mistralchatcmpl.yaml
+++ b/src/langbot/pkg/provider/modelmgr/requesters/mistralchatcmpl.yaml
@@ -1,30 +0,0 @@
 apiVersion: v1
 kind: LLMAPIRequester
 metadata:
  name: mistral-chat-completions
  label:
    en_US: Mistral AI
    zh_Hans: Mistral AI
  icon: mistral.svg
 spec:
  litellm_provider: mistral
  config:
  - name: base_url
    label:
      en_US: Base URL
      zh_Hans: 基础 URL
    type: string
    required: true
    default: https://api.mistral.ai/v1
  - name: timeout
    label:
      en_US: Timeout
      zh_Hans: 超时时间
    type: integer
    required: true
    default: 120
  support_type:
  - llm
  - text-embedding
  - rerank
  provider_category: manufacturer
--- a/src/langbot/pkg/provider/modelmgr/requesters/modelscopechatcmpl.py
+++ b/src/langbot/pkg/provider/modelmgr/requesters/modelscopechatcmpl.py
@@ -0,0 +1,561 @@
 from __future__ import annotations
 import asyncio
 import typing
 import openai
 import openai.types.chat.chat_completion as chat_completion
 import httpx
 from .. import entities, errors, requester
 import langbot_plugin.api.entities.builtin.resource.tool as resource_tool
 import langbot_plugin.api.entities.builtin.pipeline.query as pipeline_query
 import langbot_plugin.api.entities.builtin.provider.message as provider_message
 class ModelScopeChatCompletions(requester.ProviderAPIRequester):
    """ModelScope ChatCompletion API 请求器"""
    client: openai.AsyncClient
    default_config: dict[str, typing.Any] = {
        'base_url': 'https://api-inference.modelscope.cn/v1',
        'timeout': 120,
    }
    async def initialize(self):
        self.client = openai.AsyncClient(
            api_key=self.init_api_key,
            base_url=self.requester_cfg['base_url'],
            timeout=self.requester_cfg['timeout'],
            http_client=httpx.AsyncClient(trust_env=True, timeout=self.requester_cfg['timeout']),
        )
    def _mask_api_key(self, api_key: str | None) -> str:
        if not api_key:
            return ''
        if len(api_key) <= 8:
            return '****'
        return f'{api_key[:4]}...{api_key[-4:]}'
    def _infer_model_type(self, model_id: str) -> str:
        normalized_model_id = (model_id or '').lower()
        embedding_keywords = (
            'embedding',
            'embed',
            'bge-',
            'e5-',
            'm3e',
            'gte-',
            'multilingual-e5',
            'text-embedding',
        )
        return 'embedding' if any(keyword in normalized_model_id for keyword in embedding_keywords) else 'llm'
    def _infer_model_abilities(self, item: dict[str, typing.Any], model_id: str) -> list[str]:
        normalized_model_id = (model_id or '').lower()
        abilities: set[str] = set()
        def _flatten(value: typing.Any) -> list[str]:
            if value is None:
                return []
            if isinstance(value, str):
                return [value.lower()]
            if isinstance(value, dict):
                flattened: list[str] = []
                for nested_value in value.values():
                    flattened.extend(_flatten(nested_value))
                return flattened
            if isinstance(value, (list, tuple, set)):
                flattened: list[str] = []
                for nested_value in value:
                    flattened.extend(_flatten(nested_value))
                return flattened
            return [str(value).lower()]
        capability_tokens = _flatten(item.get('capabilities'))
        capability_tokens.extend(_flatten(item.get('modalities')))
        capability_tokens.extend(_flatten(item.get('input_modalities')))
        capability_tokens.extend(_flatten(item.get('output_modalities')))
        capability_tokens.extend(_flatten(item.get('supported_generation_methods')))
        capability_tokens.extend(_flatten(item.get('supported_parameters')))
        capability_tokens.extend(_flatten(item.get('architecture')))
        combined_tokens = capability_tokens + [normalized_model_id]
        vision_keywords = ('vision', 'image', 'file', 'video', 'multimodal', 'vl', 'ocr', 'omni')
        function_call_keywords = ('function', 'tool', 'tools', 'tool_choice', 'tool_call', 'tool-use', 'tool_use')
        if any(any(keyword in token for keyword in vision_keywords) for token in combined_tokens):
            abilities.add('vision')
        if any(any(keyword in token for keyword in function_call_keywords) for token in combined_tokens):
            abilities.add('func_call')
        return sorted(abilities)
    def _normalize_modalities(self, value: typing.Any) -> list[str]:
        normalized: list[str] = []
        def _collect(item: typing.Any):
            if item is None:
                return
            if isinstance(item, str):
                for part in item.replace('->', ',').replace('+', ',').split(','):
                    token = part.strip().lower()
                    if token and token not in normalized:
                        normalized.append(token)
                return
            if isinstance(item, dict):
                for nested in item.values():
                    _collect(nested)
                return
            if isinstance(item, (list, tuple, set)):
                for nested in item:
                    _collect(nested)
                return
        _collect(value)
        return normalized
    def _extract_scan_metadata(self, item: dict[str, typing.Any], model_id: str) -> dict[str, typing.Any]:
        display_name = item.get('name')
        if not isinstance(display_name, str) or not display_name.strip() or display_name == model_id:
            display_name = ''
        description = item.get('description')
        if not isinstance(description, str) or not description.strip():
            description = ''
        context_length = item.get('context_length')
        if context_length is None and isinstance(item.get('top_provider'), dict):
            context_length = item['top_provider'].get('context_length')
        if not isinstance(context_length, int):
            try:
                context_length = int(context_length) if context_length is not None else None
            except (TypeError, ValueError):
                context_length = None
        input_modalities = self._normalize_modalities(item.get('input_modalities'))
        output_modalities = self._normalize_modalities(item.get('output_modalities'))
        if isinstance(item.get('architecture'), dict):
            if not input_modalities:
                input_modalities = self._normalize_modalities(item['architecture'].get('input_modalities'))
            if not output_modalities:
                output_modalities = self._normalize_modalities(item['architecture'].get('output_modalities'))
        owned_by = item.get('owned_by')
        if not isinstance(owned_by, str) or not owned_by.strip():
            owned_by = ''
        return {
            'display_name': display_name or None,
            'description': description or None,
            'context_length': context_length,
            'owned_by': owned_by or None,
            'input_modalities': input_modalities,
            'output_modalities': output_modalities,
        }
    async def scan_models(self, api_key: str | None = None) -> dict[str, typing.Any]:
        headers = {}
        if api_key:
            headers['Authorization'] = f'Bearer {api_key}'
        models_url = f'{self.requester_cfg["base_url"].rstrip("/")}/models'
        async with httpx.AsyncClient(trust_env=True, timeout=self.requester_cfg['timeout']) as client:
            response = await client.get(models_url, headers=headers)
            response.raise_for_status()
            payload = response.json()
        models = []
        for item in payload.get('data', []):
            model_id = item.get('id')
            if not model_id:
                continue
            models.append(
                {
                    'id': model_id,
                    'name': model_id,
                    'type': self._infer_model_type(model_id),
                    'abilities': self._infer_model_abilities(item, model_id),
                    **self._extract_scan_metadata(item, model_id),
                }
            )
        models.sort(key=lambda item: (item['type'] != 'llm', item['name'].lower()))
        return {
            'models': models,
            'debug': {
                'request': {
                    'method': 'GET',
                    'url': models_url,
                    'headers': {
                        'Authorization': f'Bearer {self._mask_api_key(api_key)}' if api_key else '',
                    },
                },
                'response': payload,
            },
        }
    async def _req(
        self,
        query: pipeline_query.Query,
        args: dict,
        extra_body: dict = {},
        remove_think: bool = False,
    ) -> list[dict[str, typing.Any]]:
        args['stream'] = True
        chunk = None
        pending_content = ''
        tool_calls = []
        resp_gen: openai.AsyncStream = await self.client.chat.completions.create(**args, extra_body=extra_body)
        chunk_idx = 0
        thinking_started = False
        thinking_ended = False
        tool_id = ''
        tool_name = ''
        message_delta = {}
        async for chunk in resp_gen:
            if not chunk or not chunk.id or not chunk.choices or not chunk.choices[0] or not chunk.choices[0].delta:
                continue
            delta = chunk.choices[0].delta.model_dump() if hasattr(chunk.choices[0], 'delta') else {}
            reasoning_content = delta.get('reasoning_content')
            # 处理 reasoning_content
            if reasoning_content:
                # accumulated_reasoning += reasoning_content
                # 如果设置了 remove_think，跳过 reasoning_content
                if remove_think:
                    chunk_idx += 1
                    continue
                # 第一次出现 reasoning_content，添加 <think> 开始标签
                if not thinking_started:
                    thinking_started = True
                    pending_content += '<think>\n' + reasoning_content
                else:
                    # 继续输出 reasoning_content
                    pending_content += reasoning_content
            elif thinking_started and not thinking_ended and delta.get('content'):
                # reasoning_content 结束，normal content 开始，添加 </think> 结束标签
                thinking_ended = True
                pending_content += '\n</think>\n' + delta.get('content')
            if delta.get('content') is not None:
                pending_content += delta.get('content')
            if delta.get('tool_calls') is not None:
                for tool_call in delta.get('tool_calls'):
                    if tool_call['id'] != '':
                        tool_id = tool_call['id']
                    if tool_call['function']['name'] is not None:
                        tool_name = tool_call['function']['name']
                    if tool_call['function']['arguments'] is None:
                        continue
                    tool_call['id'] = tool_id
                    tool_call['name'] = tool_name
                    for tc in tool_calls:
                        if tc['index'] == tool_call['index']:
                            tc['function']['arguments'] += tool_call['function']['arguments']
                            break
                    else:
                        tool_calls.append(tool_call)
            if chunk.choices[0].finish_reason is not None:
                break
        message_delta['content'] = pending_content
        message_delta['role'] = 'assistant'
        message_delta['tool_calls'] = tool_calls if tool_calls else None
        return [message_delta]
    async def _make_msg(
        self,
        chat_completion: list[dict[str, typing.Any]],
    ) -> provider_message.Message:
        chatcmpl_message = chat_completion[0]
        # 确保 role 字段存在且不为 None
        if 'role' not in chatcmpl_message or chatcmpl_message['role'] is None:
            chatcmpl_message['role'] = 'assistant'
        message = provider_message.Message(**chatcmpl_message)
        return message
    async def _closure(
        self,
        query: pipeline_query.Query,
        req_messages: list[dict],
        use_model: requester.RuntimeLLMModel,
        use_funcs: list[resource_tool.LLMTool] = None,
        extra_args: dict[str, typing.Any] = {},
        remove_think: bool = False,
    ) -> tuple[provider_message.Message, dict]:
        self.client.api_key = use_model.provider.token_mgr.get_token()
        args = {}
        args['model'] = use_model.model_entity.name
        if use_funcs:
            tools = await self.ap.tool_mgr.generate_tools_for_openai(use_funcs)
            if tools:
                args['tools'] = tools
        # 设置此次请求中的messages
        messages = req_messages.copy()
        # 检查vision
        for msg in messages:
            if 'content' in msg and isinstance(msg['content'], list):
                for me in msg['content']:
                    if me['type'] == 'image_base64':
                        me['image_url'] = {'url': me['image_base64']}
                        me['type'] = 'image_url'
                        del me['image_base64']
        args['messages'] = messages
        # 发送请求
        resp = await self._req(query, args, extra_body=extra_args, remove_think=remove_think)
        # 处理请求结果
        message = await self._make_msg(resp)
        # ModelScope uses streaming, usage info not available
        usage_info = {}
        return message, usage_info
    async def _req_stream(
        self,
        args: dict,
        extra_body: dict = {},
    ) -> chat_completion.ChatCompletion:
        async for chunk in await self.client.chat.completions.create(**args, extra_body=extra_body):
            yield chunk
    async def _closure_stream(
        self,
        query: pipeline_query.Query,
        req_messages: list[dict],
        use_model: requester.RuntimeLLMModel,
        use_funcs: list[resource_tool.LLMTool] = None,
        extra_args: dict[str, typing.Any] = {},
        remove_think: bool = False,
    ) -> provider_message.Message | typing.AsyncGenerator[provider_message.MessageChunk, None]:
        self.client.api_key = use_model.provider.token_mgr.get_token()
        args = {}
        args['model'] = use_model.model_entity.name
        if use_funcs:
            tools = await self.ap.tool_mgr.generate_tools_for_openai(use_funcs)
            if tools:
                args['tools'] = tools
        # 设置此次请求中的messages
        messages = req_messages.copy()
        # 检查vision
        for msg in messages:
            if 'content' in msg and isinstance(msg['content'], list):
                for me in msg['content']:
                    if me['type'] == 'image_base64':
                        me['image_url'] = {'url': me['image_base64']}
                        me['type'] = 'image_url'
                        del me['image_base64']
        args['messages'] = messages
        args['stream'] = True
        # 流式处理状态
        # tool_calls_map: dict[str, provider_message.ToolCall] = {}
        chunk_idx = 0
        thinking_started = False
        thinking_ended = False
        role = 'assistant'  # 默认角色
        # accumulated_reasoning = ''  # 仅用于判断何时结束思维链
        async for chunk in self._req_stream(args, extra_body=extra_args):
            # 解析 chunk 数据
            if hasattr(chunk, 'choices') and chunk.choices:
                choice = chunk.choices[0]
                delta = choice.delta.model_dump() if hasattr(choice, 'delta') else {}
                finish_reason = getattr(choice, 'finish_reason', None)
            else:
                delta = {}
                finish_reason = None
            # 从第一个 chunk 获取 role，后续使用这个 role
            if 'role' in delta and delta['role']:
                role = delta['role']
            # 获取增量内容
            delta_content = delta.get('content', '')
            reasoning_content = delta.get('reasoning_content', '')
            # 处理 reasoning_content
            if reasoning_content:
                # accumulated_reasoning += reasoning_content
                # 如果设置了 remove_think，跳过 reasoning_content
                if remove_think:
                    chunk_idx += 1
                    continue
                # 第一次出现 reasoning_content，添加 <think> 开始标签
                if not thinking_started:
                    thinking_started = True
                    delta_content = '<think>\n' + reasoning_content
                else:
                    # 继续输出 reasoning_content
                    delta_content = reasoning_content
            elif thinking_started and not thinking_ended and delta_content:
                # reasoning_content 结束，normal content 开始，添加 </think> 结束标签
                thinking_ended = True
                delta_content = '\n</think>\n' + delta_content
            # 处理 content 中已有的 <think> 标签（如果需要移除）
            # if delta_content and remove_think and '<think>' in delta_content:
            #     import re
            #
            #     # 移除 <think> 标签及其内容
            #     delta_content = re.sub(r'<think>.*?</think>', '', delta_content, flags=re.DOTALL)
            # 处理工具调用增量
            if delta.get('tool_calls'):
                for tool_call in delta['tool_calls']:
                    if tool_call['id'] != '':
                        tool_id = tool_call['id']
                    if tool_call['function']['name'] is not None:
                        tool_name = tool_call['function']['name']
                    if tool_call['type'] is None:
                        tool_call['type'] = 'function'
                    tool_call['id'] = tool_id
                    tool_call['function']['name'] = tool_name
                    tool_call['function']['arguments'] = (
                        '' if tool_call['function']['arguments'] is None else tool_call['function']['arguments']
                    )
            # 跳过空的第一个 chunk（只有 role 没有内容）
            if chunk_idx == 0 and not delta_content and not reasoning_content and not delta.get('tool_calls'):
                chunk_idx += 1
                continue
            # 构建 MessageChunk - 只包含增量内容
            chunk_data = {
                'role': role,
                'content': delta_content if delta_content else None,
                'tool_calls': delta.get('tool_calls'),
                'is_final': bool(finish_reason),
            }
            # 移除 None 值
            chunk_data = {k: v for k, v in chunk_data.items() if v is not None}
            yield provider_message.MessageChunk(**chunk_data)
            chunk_idx += 1
            # return
    async def invoke_llm(
        self,
        query: pipeline_query.Query,
        model: entities.LLMModelInfo,
        messages: typing.List[provider_message.Message],
        funcs: typing.List[resource_tool.LLMTool] = None,
        extra_args: dict[str, typing.Any] = {},
        remove_think: bool = False,
    ) -> provider_message.Message:
        req_messages = []  # req_messages 仅用于类内，外部同步由 query.messages 进行
        for m in messages:
            msg_dict = m.dict(exclude_none=True)
            content = msg_dict.get('content')
            if isinstance(content, list):
                # 检查 content 列表中是否每个部分都是文本
                if all(isinstance(part, dict) and part.get('type') == 'text' for part in content):
                    # 将所有文本部分合并为一个字符串
                    msg_dict['content'] = '\n'.join(part['text'] for part in content)
            req_messages.append(msg_dict)
        try:
            return await self._closure(
                query=query,
                req_messages=req_messages,
                use_model=model,
                use_funcs=funcs,
                extra_args=extra_args,
                remove_think=remove_think,
            )
        except asyncio.TimeoutError:
            raise errors.RequesterError('请求超时')
        except openai.BadRequestError as e:
            if 'context_length_exceeded' in e.message:
                raise errors.RequesterError(f'上文过长，请重置会话: {e.message}')
            else:
                raise errors.RequesterError(f'请求参数错误: {e.message}')
        except openai.AuthenticationError as e:
            raise errors.RequesterError(f'无效的 api-key: {e.message}')
        except openai.NotFoundError as e:
            raise errors.RequesterError(f'请求路径错误: {e.message}')
        except openai.RateLimitError as e:
            raise errors.RequesterError(f'请求过于频繁或余额不足: {e.message}')
        except openai.APIError as e:
            raise errors.RequesterError(f'请求错误: {e.message}')
    async def invoke_llm_stream(
        self,
        query: pipeline_query.Query,
        model: requester.RuntimeLLMModel,
        messages: typing.List[provider_message.Message],
        funcs: typing.List[resource_tool.LLMTool] = None,
        extra_args: dict[str, typing.Any] = {},
        remove_think: bool = False,
    ) -> provider_message.MessageChunk:
        req_messages = []  # req_messages 仅用于类内，外部同步由 query.messages 进行
        for m in messages:
            msg_dict = m.dict(exclude_none=True)
            content = msg_dict.get('content')
            if isinstance(content, list):
                # 检查 content 列表中是否每个部分都是文本
                if all(isinstance(part, dict) and part.get('type') == 'text' for part in content):
                    # 将所有文本部分合并为一个字符串
                    msg_dict['content'] = '\n'.join(part['text'] for part in content)
            req_messages.append(msg_dict)
        try:
            async for item in self._closure_stream(
                query=query,
                req_messages=req_messages,
                use_model=model,
                use_funcs=funcs,
                extra_args=extra_args,
                remove_think=remove_think,
            ):
                yield item
        except asyncio.TimeoutError:
            raise errors.RequesterError('请求超时')
        except openai.BadRequestError as e:
            if 'context_length_exceeded' in e.message:
                raise errors.RequesterError(f'上文过长，请重置会话: {e.message}')
            else:
                raise errors.RequesterError(f'请求参数错误: {e.message}')
        except openai.AuthenticationError as e:
            raise errors.RequesterError(f'无效的 api-key: {e.message}')
        except openai.NotFoundError as e:
            raise errors.RequesterError(f'请求路径错误: {e.message}')
        except openai.RateLimitError as e:
            raise errors.RequesterError(f'请求过于频繁或余额不足: {e.message}')
        except openai.APIError as e:
            raise errors.RequesterError(f'请求错误: {e.message}')
--- a/src/langbot/pkg/provider/modelmgr/requesters/modelscopechatcmpl.yaml
+++ b/src/langbot/pkg/provider/modelmgr/requesters/modelscopechatcmpl.yaml
@@ -7,7 +7,6 @@ metadata:
    zh_Hans: 魔搭社区
  icon: modelscope.svg
 spec:
  litellm_provider: openai
  config:
  - name: base_url
    label:
@@ -32,8 +31,6 @@ spec:
    default: 120
  support_type:
  - llm
  - text-embedding
  - rerank
  provider_category: maas
 execution:
  python:
--- a/src/langbot/pkg/provider/modelmgr/requesters/moonshotchatcmpl.py
+++ b/src/langbot/pkg/provider/modelmgr/requesters/moonshotchatcmpl.py
@@ -0,0 +1,67 @@
 from __future__ import annotations
 import typing
 from . import chatcmpl
 from .. import requester
 import langbot_plugin.api.entities.builtin.resource.tool as resource_tool
 import langbot_plugin.api.entities.builtin.pipeline.query as pipeline_query
 import langbot_plugin.api.entities.builtin.provider.message as provider_message
 class MoonshotChatCompletions(chatcmpl.OpenAIChatCompletions):
    """Moonshot ChatCompletion API 请求器"""
    default_config: dict[str, typing.Any] = {
        'base_url': 'https://api.moonshot.cn/v1',
        'timeout': 120,
    }
    async def _closure(
        self,
        query: pipeline_query.Query,
        req_messages: list[dict],
        use_model: requester.RuntimeLLMModel,
        use_funcs: list[resource_tool.LLMTool] = None,
        extra_args: dict[str, typing.Any] = {},
        remove_think: bool = False,
    ) -> tuple[provider_message.Message, dict]:
        self.client.api_key = use_model.provider.token_mgr.get_token()
        args = {}
        args['model'] = use_model.model_entity.name
        if use_funcs:
            tools = await self.ap.tool_mgr.generate_tools_for_openai(use_funcs)
            if tools:
                args['tools'] = tools
        # 设置此次请求中的messages
        messages = req_messages
        # deepseek 不支持多模态，把content都转换成纯文字
        for m in messages:
            if 'content' in m and isinstance(m['content'], list):
                m['content'] = ' '.join([c['text'] for c in m['content']])
        # 删除空的，不知道干嘛的，直接删了。
        # messages = [m for m in messages if m["content"].strip() != "" and ('tool_calls' not in m or not m['tool_calls'])]
        args['messages'] = messages
        # 发送请求
        resp = await self._req(args, extra_body=extra_args)
        # 处理请求结果
        message = await self._make_msg(resp, remove_think)
        # Extract token usage from response
        usage_info = {}
        if hasattr(resp, 'usage') and resp.usage:
            usage_info['input_tokens'] = resp.usage.prompt_tokens or 0
            usage_info['output_tokens'] = resp.usage.completion_tokens or 0
            usage_info['total_tokens'] = resp.usage.total_tokens or 0
        return message, usage_info
--- a/src/langbot/pkg/provider/modelmgr/requesters/moonshotchatcmpl.yaml
+++ b/src/langbot/pkg/provider/modelmgr/requesters/moonshotchatcmpl.yaml
@@ -7,7 +7,6 @@ metadata:
    zh_Hans: 月之暗面
  icon: moonshot.png
 spec:
  litellm_provider: openai
  config:
  - name: base_url
    label:
@@ -25,8 +24,6 @@ spec:
    default: 120
  support_type:
  - llm
  - text-embedding
  - rerank
  provider_category: manufacturer
 execution:
  python:
--- a/src/langbot/pkg/provider/modelmgr/requesters/newapichatcmpl.py
+++ b/src/langbot/pkg/provider/modelmgr/requesters/newapichatcmpl.py
@@ -0,0 +1,17 @@
 from __future__ import annotations
 import typing
 import openai
 from . import chatcmpl
 class NewAPIChatCompletions(chatcmpl.OpenAIChatCompletions):
    """New API ChatCompletion API 请求器"""
    client: openai.AsyncClient
    default_config: dict[str, typing.Any] = {
        'base_url': 'http://localhost:3000/v1',
        'timeout': 120,
    }
--- a/src/langbot/pkg/provider/modelmgr/requesters/newapichatcmpl.yaml
+++ b/src/langbot/pkg/provider/modelmgr/requesters/newapichatcmpl.yaml
@@ -7,7 +7,6 @@ metadata:
    zh_Hans: New API
  icon: newapi.png
 spec:
  litellm_provider: openai
  config:
  - name: base_url
    label:
--- a/src/langbot/pkg/provider/modelmgr/requesters/ollamachat.py
+++ b/src/langbot/pkg/provider/modelmgr/requesters/ollamachat.py
@@ -0,0 +1,314 @@
 from __future__ import annotations
 import asyncio
 import os
 import typing
 from typing import Union, Mapping, Any, AsyncIterator
 import uuid
 import json
 import ollama
 import httpx
 from .. import errors, requester
 import langbot_plugin.api.entities.builtin.resource.tool as resource_tool
 import langbot_plugin.api.entities.builtin.pipeline.query as pipeline_query
 import langbot_plugin.api.entities.builtin.provider.message as provider_message
 REQUESTER_NAME: str = 'ollama-chat'
 class OllamaChatCompletions(requester.ProviderAPIRequester):
    """Ollama平台 ChatCompletion API请求器"""
    client: ollama.AsyncClient
    default_config: dict[str, typing.Any] = {
        'base_url': 'http://127.0.0.1:11434',
        'timeout': 120,
    }
    async def initialize(self):
        os.environ['OLLAMA_HOST'] = self.requester_cfg['base_url']
        self.client = ollama.AsyncClient(timeout=self.requester_cfg['timeout'])
    def _infer_model_type(self, model_id: str) -> str:
        normalized_model_id = (model_id or '').lower()
        embedding_keywords = ('embedding', 'embed', 'bge-', 'e5-', 'm3e', 'gte-', 'text-embedding')
        return 'embedding' if any(keyword in normalized_model_id for keyword in embedding_keywords) else 'llm'
    def _infer_model_abilities(self, item: dict[str, typing.Any], model_id: str) -> list[str]:
        normalized_model_id = (model_id or '').lower()
        abilities: set[str] = set()
        details = item.get('details', {}) or {}
        families = details.get('families', []) or []
        tokens = [normalized_model_id, str(details.get('family', '')).lower()]
        tokens.extend(str(family).lower() for family in families)
        if any(keyword in token for token in tokens for keyword in ('vision', 'vl', 'omni', 'llava', 'ocr')):
            abilities.add('vision')
        if any(keyword in token for token in tokens for keyword in ('tool', 'function')):
            abilities.add('func_call')
        return sorted(abilities)
    async def scan_models(self, api_key: str | None = None) -> dict[str, typing.Any]:
        del api_key
        models_url = f'{self.requester_cfg["base_url"].rstrip("/")}/api/tags'
        async with httpx.AsyncClient(trust_env=True, timeout=self.requester_cfg['timeout']) as client:
            response = await client.get(models_url)
            response.raise_for_status()
            payload = response.json()
        models: list[dict[str, typing.Any]] = []
        for item in payload.get('models', []):
            model_id = item.get('model') or item.get('name')
            if not model_id:
                continue
            models.append(
                {
                    'id': model_id,
                    'name': item.get('name', model_id),
                    'type': self._infer_model_type(model_id),
                    'abilities': self._infer_model_abilities(item, model_id),
                }
            )
        models.sort(key=lambda item: (item['type'] != 'llm', item['name'].lower()))
        return {
            'models': models,
            'debug': {
                'request': {
                    'method': 'GET',
                    'url': models_url,
                },
                'response': payload,
            },
        }
    async def _req(
        self,
        args: dict,
    ) -> Union[Mapping[str, Any], AsyncIterator[Mapping[str, Any]]]:
        return await self.client.chat(**args)
    async def _closure(
        self,
        query: pipeline_query.Query,
        req_messages: list[dict],
        use_model: requester.RuntimeLLMModel,
        use_funcs: list[resource_tool.LLMTool] = None,
        extra_args: dict[str, typing.Any] = {},
        remove_think: bool = False,
    ) -> provider_message.Message:
        args = extra_args.copy()
        args['model'] = use_model.model_entity.name
        messages: list[dict] = req_messages.copy()
        for msg in messages:
            if 'content' in msg and isinstance(msg['content'], list):
                text_content: list = []
                image_urls: list = []
                for me in msg['content']:
                    if me['type'] == 'text':
                        text_content.append(me['text'])
                    elif me['type'] == 'image_base64':
                        image_urls.append(me['image_base64'])
                msg['content'] = '\n'.join(text_content)
                msg['images'] = [url.split(',')[1] for url in image_urls]
            if 'tool_calls' in msg:  # LangBot 内部以 str 存储 tool_calls 的参数，这里需要转换为 dict
                for tool_call in msg['tool_calls']:
                    tool_call['function']['arguments'] = json.loads(tool_call['function']['arguments'])
        args['messages'] = messages
        args['tools'] = []
        if use_funcs:
            tools = await self.ap.tool_mgr.generate_tools_for_openai(use_funcs)
            if tools:
                args['tools'] = tools
        resp = await self._req(args)
        message: provider_message.Message = await self._make_msg(resp)
        return message
    async def _make_msg(self, chat_completions: ollama.ChatResponse) -> provider_message.Message:
        message: ollama.Message = chat_completions.message
        if message is None:
            raise ValueError("chat_completions must contain a 'message' field")
        ret_msg: provider_message.Message = None
        if message.content is not None:
            ret_msg = provider_message.Message(role='assistant', content=message.content)
        if message.tool_calls is not None and len(message.tool_calls) > 0:
            tool_calls: list[provider_message.ToolCall] = []
            for tool_call in message.tool_calls:
                tool_calls.append(
                    provider_message.ToolCall(
                        id=uuid.uuid4().hex,
                        type='function',
                        function=provider_message.FunctionCall(
                            name=tool_call.function.name,
                            arguments=json.dumps(tool_call.function.arguments),
                        ),
                    )
                )
            ret_msg.tool_calls = tool_calls
        return ret_msg
    async def _prepare_messages(
        self,
        messages: typing.List[provider_message.Message],
    ) -> list[dict]:
        """Prepare messages for Ollama API request."""
        req_messages: list = []
        for m in messages:
            msg_dict: dict = m.dict(exclude_none=True)
            content: Any = msg_dict.get('content')
            if isinstance(content, list):
                if all(isinstance(part, dict) and part.get('type') == 'text' for part in content):
                    msg_dict['content'] = '\n'.join(part['text'] for part in content)
            req_messages.append(msg_dict)
        return req_messages
    async def invoke_llm(
        self,
        query: pipeline_query.Query,
        model: requester.RuntimeLLMModel,
        messages: typing.List[provider_message.Message],
        funcs: typing.List[resource_tool.LLMTool] = None,
        extra_args: dict[str, typing.Any] = {},
        remove_think: bool = False,
    ) -> provider_message.Message:
        req_messages = await self._prepare_messages(messages)
        try:
            return await self._closure(
                query=query,
                req_messages=req_messages,
                use_model=model,
                use_funcs=funcs,
                extra_args=extra_args,
                remove_think=remove_think,
            )
        except asyncio.TimeoutError:
            raise errors.RequesterError('请求超时')
    async def invoke_llm_stream(
        self,
        query: pipeline_query.Query,
        model: requester.RuntimeLLMModel,
        messages: typing.List[provider_message.Message],
        funcs: typing.List[resource_tool.LLMTool] = None,
        extra_args: dict[str, typing.Any] = {},
        remove_think: bool = False,
    ) -> provider_message.MessageChunk:
        req_messages = await self._prepare_messages(messages)
        try:
            args = extra_args.copy()
            args['model'] = model.model_entity.name
            # Process messages for Ollama format
            msgs: list[dict] = req_messages.copy()
            for msg in msgs:
                if 'content' in msg and isinstance(msg['content'], list):
                    text_content: list = []
                    image_urls: list = []
                    for me in msg['content']:
                        if me['type'] == 'text':
                            text_content.append(me['text'])
                        elif me['type'] == 'image_base64':
                            image_urls.append(me['image_base64'])
                    msg['content'] = '\n'.join(text_content)
                    msg['images'] = [url.split(',')[1] for url in image_urls]
                if 'tool_calls' in msg:
                    for tool_call in msg['tool_calls']:
                        tool_call['function']['arguments'] = json.loads(tool_call['function']['arguments'])
            args['messages'] = msgs
            args['tools'] = []
            if funcs:
                tools = await self.ap.tool_mgr.generate_tools_for_openai(funcs)
                if tools:
                    args['tools'] = tools
            args['stream'] = True
            chunk_idx = 0
            thinking_started = False
            thinking_ended = False
            role = 'assistant'
            async for chunk in await self.client.chat(**args):
                message: ollama.Message = chunk.message
                done = chunk.done
                delta_content = message.content or ''
                reasoning_content = getattr(message, 'thinking', '') or ''
                # Handle reasoning/thinking content
                if reasoning_content:
                    if remove_think:
                        chunk_idx += 1
                        continue
                    if not thinking_started:
                        thinking_started = True
                        delta_content = '<think>\n' + reasoning_content
                    else:
                        delta_content = reasoning_content
                elif thinking_started and not thinking_ended and delta_content:
                    thinking_ended = True
                    delta_content = '\n</think>\n' + delta_content
                # Handle tool calls
                tool_calls_data = None
                if message.tool_calls:
                    tool_calls_data = []
                    for tc in message.tool_calls:
                        tool_calls_data.append(
                            {
                                'id': uuid.uuid4().hex,
                                'type': 'function',
                                'function': {
                                    'name': tc.function.name,
                                    'arguments': json.dumps(tc.function.arguments),
                                },
                            }
                        )
                # Skip empty first chunk
                if chunk_idx == 0 and not delta_content and not reasoning_content and not tool_calls_data:
                    chunk_idx += 1
                    continue
                chunk_data = {
                    'role': role,
                    'content': delta_content if delta_content else None,
                    'tool_calls': tool_calls_data,
                    'is_final': bool(done),
                }
                chunk_data = {k: v for k, v in chunk_data.items() if v is not None}
                yield provider_message.MessageChunk(**chunk_data)
                chunk_idx += 1
        except asyncio.TimeoutError:
            raise errors.RequesterError('请求超时')
    async def invoke_embedding(
        self,
        model: requester.RuntimeEmbeddingModel,
        input_text: list[str],
        extra_args: dict[str, typing.Any] = {},
    ) -> list[list[float]]:
        return (
            await self.client.embed(
                model=model.model_entity.name,
                input=input_text,
                **extra_args,
            )
        ).embeddings
--- a/src/langbot/pkg/provider/modelmgr/requesters/ollamachat.yaml
+++ b/src/langbot/pkg/provider/modelmgr/requesters/ollamachat.yaml
@@ -7,7 +7,6 @@ metadata:
    zh_Hans: Ollama
  icon: ollama.svg
 spec:
  litellm_provider: ollama
  config:
  - name: base_url
    label:
--- a/src/langbot/pkg/provider/modelmgr/requesters/openrouterchatcmpl.py
+++ b/src/langbot/pkg/provider/modelmgr/requesters/openrouterchatcmpl.py
@@ -0,0 +1,25 @@
 from __future__ import annotations
 import typing
 import openai
 from . import modelscopechatcmpl
 class OpenRouterChatCompletions(modelscopechatcmpl.ModelScopeChatCompletions):
    """OpenRouter ChatCompletion API 请求器"""
    client: openai.AsyncClient
    default_config: dict[str, typing.Any] = {
        'base_url': 'https://openrouter.ai/api/v1',
        'timeout': 120,
    }
    async def scan_models(self, api_key: str | None = None) -> dict[str, typing.Any]:
        original_base_url = self.requester_cfg.get('base_url', '')
        self.requester_cfg['base_url'] = 'https://openrouter.ai/api/v1'
        try:
            return await super().scan_models(api_key)
        finally:
            self.requester_cfg['base_url'] = original_base_url
--- a/src/langbot/pkg/provider/modelmgr/requesters/openrouterchatcmpl.yaml
+++ b/src/langbot/pkg/provider/modelmgr/requesters/openrouterchatcmpl.yaml
@@ -7,7 +7,6 @@ metadata:
    zh_Hans: OpenRouter
  icon: openrouter.svg
 spec:
  litellm_provider: openai
  config:
  - name: base_url
    label:
--- a/src/langbot/pkg/provider/modelmgr/requesters/ppiochatcmpl.py
+++ b/src/langbot/pkg/provider/modelmgr/requesters/ppiochatcmpl.py
@@ -0,0 +1,208 @@
 from __future__ import annotations
 import openai
 import typing
 from . import chatcmpl
 from .. import requester
 import openai.types.chat.chat_completion as chat_completion
 import re
 import langbot_plugin.api.entities.builtin.provider.message as provider_message
 import langbot_plugin.api.entities.builtin.pipeline.query as pipeline_query
 import langbot_plugin.api.entities.builtin.resource.tool as resource_tool
 class PPIOChatCompletions(chatcmpl.OpenAIChatCompletions):
    """欧派云 ChatCompletion API 请求器"""
    client: openai.AsyncClient
    default_config: dict[str, typing.Any] = {
        'base_url': 'https://api.ppinfra.com/v3/openai',
        'timeout': 120,
    }
    is_think: bool = False
    async def _make_msg(
        self,
        chat_completion: chat_completion.ChatCompletion,
        remove_think: bool,
    ) -> provider_message.Message:
        chatcmpl_message = chat_completion.choices[0].message.model_dump()
        # print(chatcmpl_message.keys(), chatcmpl_message.values())
        # 确保 role 字段存在且不为 None
        if 'role' not in chatcmpl_message or chatcmpl_message['role'] is None:
            chatcmpl_message['role'] = 'assistant'
        reasoning_content = chatcmpl_message['reasoning_content'] if 'reasoning_content' in chatcmpl_message else None
        # deepseek的reasoner模型
        chatcmpl_message['content'] = await self._process_thinking_content(
            chatcmpl_message['content'], reasoning_content, remove_think
        )
        # 移除 reasoning_content 字段，避免传递给 Message
        if 'reasoning_content' in chatcmpl_message:
            del chatcmpl_message['reasoning_content']
        message = provider_message.Message(**chatcmpl_message)
        return message
    async def _process_thinking_content(
        self,
        content: str,
        reasoning_content: str = None,
        remove_think: bool = False,
    ) -> tuple[str, str]:
        """处理思维链内容
        Args:
            content: 原始内容
            reasoning_content: reasoning_content 字段内容
            remove_think: 是否移除思维链
        Returns:
            处理后的内容
        """
        if remove_think:
            content = re.sub(r'<think>.*?</think>', '', content, flags=re.DOTALL)
        else:
            if reasoning_content is not None:
                content = '<think>\n' + reasoning_content + '\n</think>\n' + content
        return content
    async def _make_msg_chunk(
        self,
        delta: dict[str, typing.Any],
        idx: int,
    ) -> provider_message.MessageChunk:
        # 处理流式chunk和完整响应的差异
        # print(chat_completion.choices[0])
        # 确保 role 字段存在且不为 None
        if 'role' not in delta or delta['role'] is None:
            delta['role'] = 'assistant'
        reasoning_content = delta['reasoning_content'] if 'reasoning_content' in delta else None
        delta['content'] = '' if delta['content'] is None else delta['content']
        # print(reasoning_content)
        # deepseek的reasoner模型
        if reasoning_content is not None:
            delta['content'] += reasoning_content
        message = provider_message.MessageChunk(**delta)
        return message
    async def _closure_stream(
        self,
        query: pipeline_query.Query,
        req_messages: list[dict],
        use_model: requester.RuntimeLLMModel,
        use_funcs: list[resource_tool.LLMTool] = None,
        extra_args: dict[str, typing.Any] = {},
        remove_think: bool = False,
    ) -> provider_message.Message | typing.AsyncGenerator[provider_message.MessageChunk, None]:
        self.client.api_key = use_model.provider.token_mgr.get_token()
        args = {}
        args['model'] = use_model.model_entity.name
        if use_funcs:
            tools = await self.ap.tool_mgr.generate_tools_for_openai(use_funcs)
            if tools:
                args['tools'] = tools
        # 设置此次请求中的messages
        messages = req_messages.copy()
        # 检查vision
        for msg in messages:
            if 'content' in msg and isinstance(msg['content'], list):
                for me in msg['content']:
                    if me['type'] == 'image_base64':
                        me['image_url'] = {'url': me['image_base64']}
                        me['type'] = 'image_url'
                        del me['image_base64']
        args['messages'] = messages
        args['stream'] = True
        # tool_calls_map: dict[str, provider_message.ToolCall] = {}
        chunk_idx = 0
        thinking_started = False
        thinking_ended = False
        role = 'assistant'  # 默认角色
        async for chunk in self._req_stream(args, extra_body=extra_args):
            # 解析 chunk 数据
            if hasattr(chunk, 'choices') and chunk.choices:
                choice = chunk.choices[0]
                delta = choice.delta.model_dump() if hasattr(choice, 'delta') else {}
                finish_reason = getattr(choice, 'finish_reason', None)
            else:
                delta = {}
                finish_reason = None
            # 从第一个 chunk 获取 role，后续使用这个 role
            if 'role' in delta and delta['role']:
                role = delta['role']
            # 获取增量内容
            delta_content = delta.get('content', '')
            # reasoning_content = delta.get('reasoning_content', '')
            if remove_think:
                if delta['content'] is not None:
                    if '<think>' in delta['content'] and not thinking_started and not thinking_ended:
                        thinking_started = True
                        continue
                    elif delta['content'] == r'</think>' and not thinking_ended:
                        thinking_ended = True
                        continue
                    elif thinking_ended and delta['content'] == '\n\n' and thinking_started:
                        thinking_started = False
                        continue
                    elif thinking_started and not thinking_ended:
                        continue
            # delta_tool_calls = None
            if delta.get('tool_calls'):
                for tool_call in delta['tool_calls']:
                    if tool_call['id'] and tool_call['function']['name']:
                        tool_id = tool_call['id']
                        tool_name = tool_call['function']['name']
                    if tool_call['id'] is None:
                        tool_call['id'] = tool_id
                    if tool_call['function']['name'] is None:
                        tool_call['function']['name'] = tool_name
                    if tool_call['function']['arguments'] is None:
                        tool_call['function']['arguments'] = ''
                    if tool_call['type'] is None:
                        tool_call['type'] = 'function'
            # 跳过空的第一个 chunk（只有 role 没有内容）
            if chunk_idx == 0 and not delta_content and not delta.get('tool_calls'):
                chunk_idx += 1
                continue
            # 构建 MessageChunk - 只包含增量内容
            chunk_data = {
                'role': role,
                'content': delta_content if delta_content else None,
                'tool_calls': delta.get('tool_calls'),
                'is_final': bool(finish_reason),
            }
            # 移除 None 值
            chunk_data = {k: v for k, v in chunk_data.items() if v is not None}
            yield provider_message.MessageChunk(**chunk_data)
            chunk_idx += 1
--- a/src/langbot/pkg/provider/modelmgr/requesters/ppiochatcmpl.yaml
+++ b/src/langbot/pkg/provider/modelmgr/requesters/ppiochatcmpl.yaml
@@ -7,7 +7,6 @@ metadata:
    zh_Hans: 派欧云
  icon: ppio.svg
 spec:
  litellm_provider: openai
  config:
  - name: base_url
    label:
--- a/src/langbot/pkg/provider/modelmgr/requesters/qhaigcchatcmpl.py
+++ b/src/langbot/pkg/provider/modelmgr/requesters/qhaigcchatcmpl.py
@@ -0,0 +1,17 @@
 from __future__ import annotations
 import openai
 import typing
 from . import chatcmpl
 class QHAIGCChatCompletions(chatcmpl.OpenAIChatCompletions):
    """启航 AI ChatCompletion API 请求器"""
    client: openai.AsyncClient
    default_config: dict[str, typing.Any] = {
        'base_url': 'https://api.qhaigc.com/v1',
        'timeout': 120,
    }
--- a/src/langbot/pkg/provider/modelmgr/requesters/qhaigcchatcmpl.yaml
+++ b/src/langbot/pkg/provider/modelmgr/requesters/qhaigcchatcmpl.yaml
@@ -7,7 +7,6 @@ metadata:
    zh_Hans: 启航 AI
  icon: qhaigc.png
 spec:
  litellm_provider: openai
  config:
  - name: base_url
    label:
--- a/Show More
+++ b/Show More
Author	SHA1	Message	Date
huanghuoguoguo	8275cfd140	fix(api): avoid mutating bot update payload	2026-05-16 10:54:04 +08:00
RockChinQ	b251fc4b89	fix(plugin): resolve plugin page asset origin	2026-05-14 15:39:17 +08:00
Junyan Qin	075c85e2bc	chore: bump version 4.9.7	2026-05-12 23:48:52 +08:00
Junyan Qin	62b63ca2ca	chore: bump langbot plugin to 0.3.11	2026-05-12 23:47:35 +08:00
fdc310	3680a80248	feat(lark): implement message sending functionality in LarkAdapter	2026-05-12 18:28:34 +08:00
fdc310	6713b57d01	feat: enhance API key normalization and improve Space OAuth callback handling	2026-05-11 15:03:30 +08:00
fdc310	ea13ef87f2	feat(provider): add API key normalization and update OpenAI requester initialization	2026-05-11 14:21:42 +08:00
fdc310	59bd581e88	feat(i18n): add 'recommend' and 'start' keys for Spanish, Russian, Thai, and Vietnamese locales	2026-05-11 10:31:32 +08:00
fdc310	cba83a62e8	feat(i18n): add Feishu, WeChat, DingTalk, and WeCombot support in multiple languages	2026-05-11 10:08:16 +08:00
Dongchuan Fu	f412127fb0	feat: add one-click app creation for Feishu/dingding/wexin/wecombot with QR code support (#2165 ) * feat: add one-click app creation for Feishu with QR code support * feat: implement WeChat QR code login functionality and update related configurations * feat: add qrcode dependency for QR code generation support * feat: enhance QR code login UI and add internationalization support for new labels * feat: new ui back * feat: add DingTalk one-click app creation and QR code login support * feat: add WeComBot one-click creation support and enhance QR code login functionality * feat: Update the robot creation function and bind the most recently updated pipeline	2026-05-10 22:31:31 +08:00
huanghuoguoguo	5273bbb23f	feat(i18n): add missing i18n keys for knowledge validation messages Add engineSettingsInvalid and retrievalSettingsInvalid keys to all locale files (zh-Hant, ja-JP, vi-VN, es-ES, ru-RU, th-TH) for the new dynamic form validation feature. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>	2026-05-10 18:29:22 +08:00
huanghuoguoguo	0ceab3f6a5	feat(knowledge): validate required fields based on plugin schema Add business-agnostic validation for knowledge base creation: - Backend: dynamically validate required fields from plugin's creation_schema and retrieval_schema, with support for show_if conditional fields - Frontend: expose validation function from DynamicFormComponent and validate before KBForm submission - Add i18n translations for validation error messages Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>	2026-05-10 18:25:28 +08:00
RockChinQ	aedc097188	fix(plugin): update runtime PyPI index defaults	2026-05-09 15:26:53 +08:00
RockChinQ	18b27dd9ef	fix(plugin): use runtime dependency failure fix	2026-05-09 14:56:56 +08:00
RockChinQ	3f50a56623	fix(plugin): surface dependency install failures	2026-05-09 14:42:05 +08:00
Junyan Chin	1fcdbd472f	fix model runtime uuid after updates (#2160 ) * fix model runtime uuid after updates * test: avoid local agent constructor coupling	2026-05-02 21:27:34 +08:00
Haoxuan Xing	547006cb4a	feat: add supports for Matrix protocol(#2110 ) * Optimize the plugin system * feat: enhance plugin installation process and improve task management * fix: linter err * feat: add Matrix adapter with multi-bridge support - MatrixAdapter with text/image/file message support - Multi-bridge architecture (BridgeState) for Discord, Telegram, etc. - Auto-login, QR forwarding, disconnect detection - Force logout+login on adapter start - Group/private chat detection excluding bridge bots - matrix-nio dependency added * docs: sync platform tables across all READMEs with Matrix bridge support - Add Matrix/Satori compatibility notes to all platforms - Add 21 Matrix-only platforms (Signal, WhatsApp, Messenger, etc.) - Keep international market ordering (Discord first) for non-CN READMEs * Update API base URL to localhost * fix: remove unused datetime import (ruff) * style: ruff format matrix.py * docs: collapse matrix platform list * docs: simplify platform compatibility notes --------- Co-authored-by: Junyan Qin <rockchinq@gmail.com>	2026-05-02 21:04:49 +08:00
Junyan Qin	92bf9a7ea5	style: make wizard steps blue	2026-05-02 18:42:34 +08:00
Junyan Qin	832efb4069	fix: hide normal storage section status badge	2026-05-02 17:38:40 +08:00
Junyan Qin	8f1847d480	fix: allow storage analysis dialog scrolling	2026-05-02 17:36:10 +08:00
Junyan Qin	fe619e415f	fix: move storage analysis to account menu	2026-05-02 17:31:09 +08:00
Junyan Chin	0154ea6cd3	Fix/storage retention cleanup (#2159 ) * fix: add storage retention cleanup * fix: prune completed tasks on completion * fix: complete storage analysis i18n	2026-05-02 17:09:31 +08:00
Junyan Chin	8db55267d8	feat(models): support object type in extra parameters (#2158 ) Add 'object' as a new value type for model extra parameters so users can configure nested JSON like {"thinking": {"type": "disabled"}} required by DeepSeek-v4 non-thinking mode (refs #2157). UI: add 'Object' option to the type dropdown in ExtraArgsEditor; render a full-width JSON Textarea (resize-y, monospace) with live JSON validation. On save, JSON is parsed and rejected if not a plain object. Also make the model edit and add-model popovers scrollable: cap height at min(70vh, --radix-popover-content-available-height), stop wheel/touchmove propagation so the dialog's react-remove-scroll lock doesn't swallow events, and use overscroll-none to avoid the bottom border seam from rubber-band overscroll. i18n updated for all 8 locales. Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>	2026-05-01 20:44:17 +08:00
Bruce	b9662250a6	add conversation expire config & user query text to dingtalk card (#2147 ) * add conversation expire config * add user query text to card * fix(pipeline): move session limit to AI config * test(pipeline): cover AI session limit config * refactor(pipeline): merge session expire-time into AI runner stage Move the session validity duration field out of the standalone session-limit stage into the runner stage so it actually renders in the AI tab (the tab only shows the runner stage and the stage matching the selected runner — any other stage is filtered out). Read path, default config, metadata description, and tests updated accordingly. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com> * fix(pipeline): expire conversations from last update time * fix(n8n): sync generated conversation id into payload --------- Co-authored-by: RockChinQ <rockchinq@gmail.com> Co-authored-by: Claude Opus 4.7 (1M context) <noreply@anthropic.com>	2026-05-01 18:13:55 +08:00
fdc310	d9378c3a88	feat: Support WebSocket mode and enhance message processing capabilities (#2156 ) * feat: Support WebSocket mode and enhance message processing capabilities * feat: add steam * feat: enhance QQOfficialClient and QQOfficialAdapter with improved logging and stream context management	2026-05-01 02:33:44 +08:00
Jack Chiang	86a4d1bf0b	feat: add Qiniu provider support (#2155 ) * feat: add Qiniu provider support * feat: add Qiniu provider support --------- Co-authored-by: JiangZhuo <jiangzhuo@qiniu.com>	2026-04-29 13:52:56 +08:00
Junyan Qin	ce6e79db8e	fix(dependencies): update langbot-plugin to version 0.3.10	2026-04-26 02:18:12 +08:00
Junyan Qin	d53e2cb9a0	fix(web): prevent tab list layout shift when save button toggles visibility Use invisible class instead of conditional rendering for save buttons in bot, pipeline, and knowledge base detail pages, so the button always occupies space and the tab list position stays stable across tab switches. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-04-26 02:15:36 +08:00
sheetung	c1168745b7	Feat/web UI fixes v2 (#2152 ) * fix(web): 修复复制按钮和插件安装对话框UI问题 - 新增 clipboard.ts 工具函数支持 Clipboard API 降级 - 修复添加机器人页面 Webhook URL 复制按钮未生效 - 修复 API 集成对话框 API Key 复制按钮未生效 - 修复 Bot 会话监控用户 ID 复制按钮未生效 - 修复插件安装进度状态框横向溢出和小屏缩放问题 Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(web): improve clipboard copy with Selection API fallback Replace navigator.clipboard.writeText with Selection API + execCommand for reliable copying in non-secure contexts. Remove duplicate dialog. Fix scanProviderModels type signature to accept rerank model type. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(web): revert package-lock.json to match upstream Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(web): fix prettier formatting errors Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com> * fix(web): unify all clipboard copy to use copyToClipboard utility - Fix embed code copy button not working in non-secure contexts - Add copy animation (check icon) to embed code button via EmbedCodeField component - Replace raw navigator.clipboard calls in plugins/page.tsx and BotLogCard.tsx - Remove duplicated inline fallback implementations Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.7 <noreply@anthropic.com> Co-authored-by: Junyan Qin <rockchinq@gmail.com>	2026-04-26 01:57:54 +08:00
Copilot	69b87a0d8a	fix(pipeline): handle File messages with base64 data in preproc (#2149 ) File messages from platforms like Telegram carry base64 data with an empty url. The unconditional from_file_url(me.url) call passed an empty string downstream, causing httpx to fail with "Request URL is missing an 'http://' or 'https://' protocol" when uploading to Dify. Mirror the existing Voice handling pattern: check base64 first, fall back to url. Applied in both the main message chain and the Quote path. Closes #2079 Co-authored-by: Junyan Qin <rockchinq@gmail.com> Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-04-25 22:43:00 +08:00
Junyan Qin	6637b153f1	fix(i18n): add missing plugin page keys to all locale files Add sidebar.pluginPages, sidebar.pluginPagesTooltip, pluginPages section, and plugins.componentName.Page to es-ES, ja-JP, ru-RU, th-TH, vi-VN, zh-Hant to fix CI i18n key check. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-04-25 22:30:01 +08:00
Junyan Qin	e768fc6116	Refactor code structure for improved readability and maintainability	2026-04-25 22:23:11 +08:00
Junyan Qin	2442d3bf52	feat(web): add Page component filter to in-app marketplace Add Page toggle button with PanelTop icon to the in-app plugin marketplace component filter bar. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-04-25 21:51:40 +08:00
Junyan Qin	42d78817f4	refactor(web): remove per-page icon from PluginPageItem Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-04-25 21:46:11 +08:00
Junyan Qin	4b9f25a05d	revert(web): remove per-page icon from sidebar sub-items Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-04-25 21:44:38 +08:00
Junyan Qin	d1f0e07cc0	feat(web): render page icon emoji in sidebar sub-items Show the per-page icon (emoji from page manifest metadata.icon) in collapsible plugin page sub-items. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-04-25 21:41:58 +08:00
Junyan Qin	78e55509ae	fix(web): add Page component icon and fix label in plugin component list Add PanelTop icon for Page components in the plugin detail component list. Change zh-Hans label from '扩展页' to '页面' for consistency. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-04-25 21:38:52 +08:00
Junyan Qin	2c28635a39	fix(web): use plugin icon in sidebar, disable text selection on entries - Replace hardcoded Puzzle/LayoutDashboard icons with actual plugin icon image loaded from the plugin icon API endpoint - Add select-none to all plugin page sidebar entries to prevent accidental text selection - Add pluginIconURL to PluginPageItem data model Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-04-25 20:10:35 +08:00
Junyan Qin	5f3cecfbe2	feat(web): group plugin pages by plugin in sidebar with collapsible sections - Group pages by plugin when a plugin has multiple pages, collapse under the plugin label; single-page plugins render directly without nesting - Rename "Extension Pages" to "Plugin Pages" with tooltip explaining these are visual pages provided by installed plugins - Add pluginLabel to PluginPageItem for display Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-04-25 20:06:03 +08:00
Junyan Qin	12df9d6ee9	feat: add plugin extension pages (iframe rendering, Page SDK, security hardening, i18n) Co-Authored-By: Typer_Body <mcjiekejiemi@163.com>	2026-04-25 19:14:14 +08:00
Sebastion	195f6efeff	fix: prevent path traversal in LocalStorageProvider via key parameter (#2087 ) Add _safe_resolve() helper that uses os.path.realpath() to canonicalize the joined path and verifies it stays within LOCAL_STORAGE_PATH. All six public methods (save, load, exists, delete, size, delete_dir_recursive) now validate the key before performing any I/O. This prevents absolute-path injection (e.g. key="/etc/passwd") and relative traversal (e.g. key="../../etc/passwd") from escaping the storage root directory. CWE-22	2026-04-24 15:46:37 +08:00
fdc310	564d829e25	Feat/webpage adapter (#2135 ) * feat: add web_page_bot adapter and embed widget - Implemented a new `web_page_bot` adapter for embedding chat widgets on websites. - Created a new YAML configuration file for `web_page_bot` with necessary metadata and execution details. - Developed the `WebPageBotAdapter` class to handle message events and manage listeners. - Added a JavaScript widget for embedding the chat interface, including styles and functionality for user interaction. - Updated WebSocket handling to support the new bot adapter and manage connections. - Enhanced the bot form to include pipeline UUID and adapter configuration in the system context. - Introduced a new dynamic form item type for embed code in the form entity. * feat(embed): add feedback submission and image upload functionality to embed widget * feat(embed): add reset session endpoint for embed widget and improve WebSocket image handling * feat(widget): remove typing indicator display logic from message handling * fix(embed): security hardening for embed widget - Add UUID format validation for pipeline_uuid parameters - Add Cloudflare Turnstile integration for bot protection (optional) - Add HMAC-signed session tokens for /messages, /reset, /feedback endpoints - Sanitize error responses (remove internal exception details) - Sanitize base_url before JS injection - Fix XSS in markdown link rendering (only allow http/https protocols) - Fix XSS in image URL extraction (only allow http/https/data protocols) - Escape widget title in embed code snippet (HTML entity encoding) - Remove class-level mutable default in WebPageBotAdapter - Remove duplicate config line and console.log in widget.js - Add turnstile_site_key and turnstile_secret_key config fields * style: fix prettier formatting for chained replace calls * fix(embed): declare listeners as Pydantic field in WebPageBotAdapter The base class is a Pydantic BaseModel, so listeners must be declared as a field (with default_factory) rather than assigned in __init__. Also keep the __init__ to convert positional args to keyword args for Pydantic compatibility with botmgr's calling convention. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * refactor(embed): use bot_uuid instead of pipeline_uuid in all embed URLs Replace pipeline_uuid with bot_uuid in all user-facing embed widget URLs so internal pipeline identifiers are never exposed. The server resolves bot_uuid to the owning web_page_bot, validates it is enabled and has a pipeline bound, then routes internally using pipeline_uuid. Add a dedicated WebSocket endpoint at /api/v1/embed/<bot_uuid>/ws/connect instead of reusing the pipeline debug path. Wire WebPageBotAdapter to proxy reply_message calls through the WebSocket adapter so dashboard shows the correct adapter name while replies are still delivered. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * docs(embed): improve Turnstile config field descriptions Add guidance on where to obtain the keys (Cloudflare dashboard) and clarify that leaving them empty disables the feature. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * feat(embed): add multi-language support for embed widget Add a language selector to the web_page_bot config with 8 locales (en, zh-Hans, zh-Hant, ja, es, ru, th, vi). The backend injects the locale into widget.js which uses a built-in i18n dictionary for all user-facing strings (welcome message, placeholder, aria labels, error messages, powered-by footer). Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix(embed): use correct select option format for language selector Options must use name/label (i18n object) format, not value/label (plain string), to match the dynamic form renderer. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * style(embed): adjust footer padding and link to langbot.app Increase footer padding for more breathing room from the bottom edge. Change powered-by link from GitHub repo to langbot.app. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix(embed): ignore Enter key during IME composition Check e.isComposing before treating Enter as send, so confirming an IME candidate (e.g. Chinese/Japanese input) does not also fire the message. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * fix(embed): center bubble icon and fill entire circle Make .lb-chat-icon span fill the full bubble area so the logo image covers the circle completely without exposing the blue background. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> * feat(embed): add bubble icon presets selector Add 6 bubble icon options (LangBot logo, chat bubble, robot, headset, sparkle, message) configurable in the bot settings. Icons are inline SVGs in widget.js, selected via a config field injected by the backend. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com> --------- Co-authored-by: RockChinQ <rockchinq@gmail.com> Co-authored-by: Claude Opus 4.6 (1M context) <noreply@anthropic.com>	2026-04-24 15:36:14 +08:00
`@@ -1,3 +1,3 @@`
	`"""LangBot - Production-grade platform for building agentic IM bots"""`	`"""LangBot - Production-grade platform for building agentic IM bots"""`

	`__version__ = '4.9.6'`	`__version__ = '4.9.7'`