refactor: remove beta diagnostics collection and reporting (#2616)

src/langbot/pkg/agent/runner/interaction_manager.py

@@ -2,8 +2,6 @@
 
 from __future__ import annotations
 
-from langbot.pkg.telemetry import diagnostics
-
 import copy
 import json
 import time
@@ -94,7 +92,6 @@ class InteractionManager:
         self.ap = ap
         self.store = store or InteractionStore(ap.persistence_mgr.get_db_engine())
 
-    @diagnostics.observe('interaction', 'interaction.request', source='agent', stage='execute')
     async def handle_result(
         self,
         *,
@@ -208,7 +205,6 @@ class InteractionManager:
         except Exception as exc:
             await self.store.mark_delivery_failed(run_id, request.interaction_id, str(exc))
             raise
-        diagnostics.set_outcome('waiting', reason_code='waiting')
         return True
 
     @staticmethod
@@ -255,14 +251,12 @@ class InteractionManager:
         except pydantic.ValidationError:
             return
 
-    @diagnostics.observe('interaction', 'interaction.acknowledge', source='platform', stage='ack')
     async def acknowledge_submission(self, record: dict[str, typing.Any], adapter: typing.Any) -> None:
         """Best-effort transition of submitted controls into a read-only state."""
         delivery_result = record.get('delivery_result')
         if not isinstance(delivery_result, dict) or not self._supports_platform_api(
             adapter, INTERACTION_ACKNOWLEDGE_API
         ):
-            diagnostics.set_outcome('skipped')
             return
         try:
             delivery_result = await adapter.call_platform_api(
@@ -280,8 +274,6 @@ class InteractionManager:
                     delivery_result,
                 )
         except Exception as exc:
-            diagnostics.set_outcome('failed', reason_code='response_error')
-            diagnostics.annotate(error=exc)
             self._warning(f'Failed to acknowledge interaction submission: {exc}')
 
     async def _find_update_target(
@@ -332,7 +324,6 @@ class InteractionManager:
                     f'interaction field {field.id} has duplicate option values',
                 )
 
-    @diagnostics.observe('interaction', 'interaction.consume', source='platform', stage='dispatch')
     async def consume_callback(
         self,
         *,

src/langbot/pkg/agent/runner/orchestrator.py

@@ -10,8 +10,6 @@ from langbot_plugin.api.entities.builtin.provider import message as provider_mes
 from langbot_plugin.api.entities.builtin.pipeline import query as pipeline_query
 
 from langbot_plugin.entities.io.actions.enums import PluginToRuntimeAction
-
-from ...telemetry import diagnostics as diagnostics
 from .reply_stream import ReplyStreamSession
 from ...core import app
 from ...api.http.context import ExecutionContext
@@ -80,7 +78,6 @@ class AgentRunOrchestrator:
         self.journal = AgentRunJournal(ap)
         self._session_registry = get_session_registry()
 
-    @diagnostics.observe('run', 'runner.run', source='agent', stage='prepare')
     async def run(
         self,
         event: AgentEventEnvelope,
@@ -106,8 +103,6 @@ class AgentRunOrchestrator:
             runner_id,
             bound_plugins,
         )
-
-        diagnostics.runner_metadata(self.ap, descriptor, binding.processor_type)
         usage = 'event' if binding.processor_type == 'event_processor' else 'agent'
         if usage not in descriptor.usages:
             raise ValueError(f'The selected Runner does not support {usage} usage')
@@ -177,7 +172,6 @@ class AgentRunOrchestrator:
 
         state_context = build_state_context(event, binding, descriptor)
         run_id = context['run_id']
-        diagnostics.annotate(run_id=run_id, stage='execute')
         context['context']['available_apis']['reply_stream'] = hasattr(PluginToRuntimeAction, 'REPLY_STREAM') and any(
             tool.get('tool_name') == 'event_reply' and tool.get('tool_type') == 'platform'
             for tool in resources.get('tools', [])
@@ -188,7 +182,6 @@ class AgentRunOrchestrator:
             source=(adapter_context or {}).get('_platform_event')
             or getattr((adapter_context or {}).get('_query'), 'message_event', None),
         )
-        reply_streams.diagnostics = getattr(self.ap, 'diagnostics', None)
         available_apis = context.get('context', {}).get('available_apis')
         run_authorization = {
             'runner_id': descriptor.id,
@@ -390,12 +383,6 @@ class AgentRunOrchestrator:
                         terminal_status = 'cancelled'
                         terminal_reason = run_snapshot.get('status_reason') or 'cancel_requested'
                         break
-            diagnostics.set_outcome(
-                {'completed': 'succeeded', 'failed': 'failed', 'cancelled': 'cancelled'}.get(
-                    terminal_status, 'succeeded'
-                ),
-                reason_code='runner_failed' if terminal_status == 'failed' else '',
-            )
             await self.journal.finalize_run(
                 run_id=run_id,
                 status=terminal_status or 'completed',
@@ -403,7 +390,6 @@ class AgentRunOrchestrator:
                 usage=terminal_usage,
             )
         except Exception as exc:
-            diagnostics.set_outcome('timeout' if self._is_deadline_exhausted(context) else 'failed')
             failed_usage = terminal_usage
             await self.journal.finalize_run(
                 run_id=run_id,
@@ -432,7 +418,6 @@ class AgentRunOrchestrator:
                         exc_info=True,
                     )
 
-    @diagnostics.observe('lifecycle', 'runner.query_prepare', source='pipeline', stage='prepare')
     async def run_from_query(
         self,
         query: pipeline_query.Query,

src/langbot/pkg/agent/runner/platform_tools.py

@@ -2,8 +2,6 @@
 
 from __future__ import annotations
 
-from ...telemetry import diagnostics
-
 import copy
 import fnmatch
 import typing
@@ -669,7 +667,6 @@ def resolve_platform_api_call(session, bot_uuid, action, params, context_tool=No
     raise ValueError(f'Platform API {action} or its target is not authorized for this run')
 
 
-@diagnostics.observe('api', 'host.platform_tool', source='agent')
 async def execute_platform_tool(
     ap: typing.Any,
     execution_context: typing.Any,
@@ -689,7 +686,6 @@ async def execute_platform_tool(
         normalized = _event_params(definition, context, normalized)
     # This flag is frozen by the Host from the synthetic debug envelope, not tool arguments.
     if delivery.get('surface') == 'webui' and (delivery.get('platform_capabilities') or {}).get('debug_mock') is True:
-        diagnostics.annotate(source='webui_debug', attributes={'synthetic': True})
         result = _execute_mock_platform_tool(definition, context, normalized)
         if message_chain is not None:
             result['parameters']['message'] = message_chain.model_dump(mode='json')

src/langbot/pkg/agent/runner/registry.py

@@ -2,8 +2,6 @@
 
 from __future__ import annotations
 
-from ...telemetry import diagnostics
-
 import typing
 import asyncio
 
@@ -138,9 +136,6 @@ class RunnerRegistry:
             permissions=typed_manifest.permissions,
             raw_manifest=manifest,
         )
-        manager = getattr(self.ap, 'diagnostics', None)
-        if isinstance(manager, diagnostics.DiagnosticsManager) and manager.enabled:
-            diagnostics.declare_runner(descriptor)
         return descriptor
 
     async def refresh(self, context: TenantContext) -> None:

src/langbot/pkg/agent/runner/reply_stream.py

@@ -7,8 +7,6 @@ from dataclasses import dataclass, field
 from typing import Literal
 from uuid import UUID, uuid4
 
-from ...telemetry import diagnostics
-
 import pydantic
 from langbot_plugin.api.entities.builtin.platform import events, message
 from langbot_plugin.api.entities.builtin.provider import message as provider_message
@@ -56,9 +54,6 @@ class ReplyStreamSession:
         self._closed = False
         self._active: set[asyncio.Task] = set()
 
-    @diagnostics.observe(
-        'delivery', 'reply_stream.apply', source='agent', fields=lambda b: {'attributes': {'stream': True}}
-    )
     async def apply(self, request: ReplyStreamRequest) -> dict:
         task = asyncio.create_task(self._apply(request))
         self._active.add(task)

src/langbot/pkg/agent/runner/session_registry.py

@@ -2,8 +2,6 @@
 
 from __future__ import annotations
 
-from ...telemetry import diagnostics
-
 import asyncio
 import copy
 import typing
@@ -172,7 +170,6 @@ class AgentRunSessionRegistry:
             'query_id': query_id,
             'execution_query': execution_query,
             'reply_streams': reply_streams,
-            '_diagnostic_context': diagnostics.capture_context(),
             'plugin_identity': plugin_identity,
             'authorization': authorization,
             'status': {