* feat: centralize telemetry logging at LLM client level
Moves telemetry logging from individual adapters to LLMClientBase:
- Add TelemetryStreamWrapper for streaming telemetry on stream close
- Add request_async_with_telemetry() for non-streaming requests
- Add stream_async_with_telemetry() for streaming requests
- Add set_telemetry_context() to configure agent_id, run_id, step_id
Updates adapters and agents to use new pattern:
- LettaLLMAdapter now accepts agent_id/run_id in constructor
- Adapters call set_telemetry_context() before LLM requests
- Removes duplicate telemetry logging from adapters
- Enriches traces with agent_id, run_id, call_type metadata
🐙 Generated with [Letta Code](https://letta.com)
Co-Authored-By: Letta <noreply@letta.com>
* fix: accumulate streaming response content for telemetry
TelemetryStreamWrapper now extracts actual response data from chunks:
- Content text (concatenated from deltas)
- Tool calls (id, name, arguments)
- Model name, finish reason, usage stats
🐙 Generated with [Letta Code](https://letta.com)
Co-Authored-By: Letta <noreply@letta.com>
* refactor: move streaming telemetry to caller (option 3)
- Remove TelemetryStreamWrapper class
- Add log_provider_trace_async() helper to LLMClientBase
- stream_async_with_telemetry() now just returns raw stream
- Callers log telemetry after processing with rich interface data
Updated callers:
- summarizer.py: logs content + usage after stream processing
- letta_agent.py: logs tool_call, reasoning, model, usage
🐙 Generated with [Letta Code](https://letta.com)
Co-Authored-By: Letta <noreply@letta.com>
* fix: pass agent_id and run_id to parent adapter class
LettaLLMStreamAdapter was not passing agent_id/run_id to parent,
causing "unexpected keyword argument" errors.
🐙 Generated with [Letta Code](https://letta.com)
Co-Authored-By: Letta <noreply@letta.com>
---------
Co-authored-by: Letta <noreply@letta.com>
* feat: add provider trace backend abstraction for multi-backend telemetry
Introduces a pluggable backend system for provider traces:
- Base class with async/sync create and read interfaces
- PostgreSQL backend (existing behavior)
- ClickHouse backend (via OTEL instrumentation)
- Socket backend (writes to Unix socket for crouton sidecar)
- Factory for instantiating backends from config
Refactors TelemetryManager to use backends with support for:
- Multi-backend writes (concurrent via asyncio.gather)
- Primary backend for reads (first in config list)
- Graceful error handling per backend
Config: LETTA_TELEMETRY_PROVIDER_TRACE_BACKEND (comma-separated)
Example: "postgres,socket" for dual-write to Postgres and crouton
🐙 Generated with [Letta Code](https://letta.com)
Co-Authored-By: Letta <noreply@letta.com>
* feat: add protocol version to socket backend records
Adds PROTOCOL_VERSION constant to socket backend:
- Included in every telemetry record sent to crouton
- Must match ProtocolVersion in apps/crouton/main.go
- Enables crouton to detect and reject incompatible messages
🐙 Generated with [Letta Code](https://letta.com)
Co-Authored-By: Letta <noreply@letta.com>
* fix: remove organization_id from ProviderTraceCreate calls
The organization_id is now handled via the actor parameter in the
telemetry manager, not through ProviderTraceCreate schema. This fixes
validation errors after changing ProviderTraceCreate to inherit from
BaseProviderTrace which forbids extra fields.
🐙 Generated with [Letta Code](https://letta.com)
Co-Authored-By: Letta <noreply@letta.com>
* consolidate provider trace
* add clickhouse-connect to fix bug on main lmao
* auto generated sdk changes, and deployment details, and clikchouse prefix bug and added fields to runs trace return api
* auto generated sdk changes, and deployment details, and clikchouse prefix bug and added fields to runs trace return api
* consolidate provider trace
* consolidate provider trace bug fix
---------
Co-authored-by: Letta <noreply@letta.com>
* feat: add TypeScript tool support for E2B sandbox execution
This change implements TypeScript tool support using the same E2B path as Python tools:
- Add TypeScript execution script generator (typescript_generator.py)
- Modify E2B sandbox to detect TypeScript tools and use language='ts'
- Add npm package installation for TypeScript tool dependencies
- Add validation requiring json_schema for TypeScript tools
- Add comprehensive integration tests for TypeScript tools
TypeScript tools:
- Require explicit json_schema (no docstring parsing)
- Use JSON serialization instead of pickle for results
- Support async functions with top-level await
- Support npm package dependencies via npm_requirements field
Closes#8793
Co-authored-by: Sarah Wooders <sarahwooders@users.noreply.github.com>
* fix: disable AgentState for TypeScript tools & add letta-client injection
Based on Sarah's feedback:
1. AgentState is a legacy Python-only feature, disabled for TS tools
2. Added @letta-ai/letta-client npm package injection for TypeScript
(similar to letta_client for Python)
Changes:
- base.py: Explicitly set inject_agent_state=False for TypeScript tools
- typescript_generator.py: Inject LettaClient initialization code
- e2b_sandbox.py: Auto-install @letta-ai/letta-client for TS tools
- Added tests verifying both behaviors
🤖 Generated with [Letta Code](https://letta.com)
Co-Authored-By: Sarah Wooders <sarahwooders@users.noreply.github.com>
Co-Authored-By: Letta <noreply@letta.com>
* Update core-integration-tests.yml
* fix: convert TypeScript test fixtures to async
The OrganizationManager and UserManager no longer have sync methods,
only async variants. Updated all fixtures to use:
- create_organization_async
- create_actor_async
- create_or_update_tool_async
🤖 Generated with [Letta Code](https://letta.com)
Co-Authored-By: Letta <noreply@letta.com>
* fix: skip Python AST parsing for TypeScript tools in sandbox base
The _init_async method was calling parse_function_arguments (which uses
Python's ast.parse) before checking if the tool was TypeScript, causing
SyntaxError when running TypeScript tools.
Moved the is_typescript_tool() check to happen first, skipping Python
AST parsing entirely for TypeScript tools.
🤖 Generated with [Letta Code](https://letta.com)
Co-Authored-By: Letta <noreply@letta.com>
* letta_agent_id
* skip ast parsing for s
* add tool execution test
---------
Co-authored-by: letta-code <248085862+letta-code@users.noreply.github.com>
Co-authored-by: Sarah Wooders <sarahwooders@users.noreply.github.com>
Co-authored-by: Letta <noreply@letta.com>
Co-authored-by: Kian Jones <kian@letta.com>
feat(core): add run_id filter to internal runs endpoint
Adds the ability to filter runs by a specific run ID in the
internal runs list endpoint.
🐛 Generated with [Letta Code](https://letta.com)
Co-authored-by: Letta <noreply@letta.com>
* feat: add /v1/runs/{run_id}/trace endpoint for OTEL traces
- Add new endpoint to retrieve filtered OTEL spans for a run
- Filter to only return UI-relevant spans (agent_step, tool executions, root span, TTFT)
- Skip Postgres writes when ClickHouse is enabled for provider traces
- Add USE_CLICKHOUSE_FOR_PROVIDER_TRACES env var to helm/justfile
- Move typecheck CI to self-hosted runners
🤖 Generated with [Letta Code](https://letta.com)
Co-Authored-By: Letta <noreply@letta.com>
* fix: add missing clickhouse_provider_traces.py
The telemetry_manager.py imports ClickhouseProviderTraceReader from
this module, but the file was not included when splitting the PR.
🤖 Generated with [Letta Code](https://letta.com)
Co-Authored-By: Letta <noreply@letta.com>
* autogen
* fix: add trace.retrieve to stainless.yml for SDK generation
Adds the runs.trace.retrieve method mapping so Stainless generates
the useRunsServiceRetrieveTraceForRun hook.
🤖 Generated with [Letta Code](https://letta.com)
Co-Authored-By: Letta <noreply@letta.com>
---------
Co-authored-by: Letta <noreply@letta.com>
* feat: add conversation_id parameter to context endpoint [LET-6989]
Add optional conversation_id query parameter to retrieve_agent_context_window.
When provided, the endpoint uses messages from the specific conversation
instead of the agent's default message_ids.
👾 Generated with [Letta Code](https://letta.com)
Co-Authored-By: Letta <noreply@letta.com>
* chore: regenerate SDK after context endpoint update [LET-6989]
👾 Generated with [Letta Code](https://letta.com)
Co-Authored-By: Letta <noreply@letta.com>
* feat: add isolated blocks support for conversations
Allows conversations to have their own copies of specific memory blocks (e.g., todo_list) that override agent defaults, enabling conversation-specific state isolation.
👾 Generated with [Letta Code](https://letta.com)
Co-Authored-By: Letta <noreply@letta.com>
* undo
* update apis
* test
* cleanup
* fix tests
* simplify
* move override logic
* patch
---------
Co-authored-by: Letta <noreply@letta.com>
This commit addresses the httpx.ReadTimeout error detected in production
by adding explicit timeout configurations to several httpx client usages:
1. MCP SSE client: Pass mcp_connect_to_server_timeout (30s) to sse_client()
2. MCP StreamableHTTP client: Pass mcp_connect_to_server_timeout (30s) to streamablehttp_client()
3. OpenAI model list API: Add 30s timeout with 10s connect timeout
4. Google AI model list/details API: Add 30s timeout with 10s connect timeout
Previously, these httpx clients were created without explicit timeouts,
which could cause ReadTimeout errors when remote servers are slow to respond.
Fixes#8073🤖 Generated with [Letta Code](https://letta.com)
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: datadog-official[bot] <datadog-official[bot]@users.noreply.github.com>
Co-authored-by: Kian Jones <11655409+kianjones9@users.noreply.github.com>
The FastMCP clients were not properly wrapping exceptions from
`connect_to_server()`, causing raw RuntimeErrors (like DNS resolution
failures with "[Errno -2] Name or service not known") to propagate
up unchanged.
Changes:
- Both `AsyncFastMCPSSEClient` and `AsyncFastMCPStreamableHTTPClient`
now properly catch all exceptions and wrap them in `ConnectionError`
- Added warning-level logging for failed connections
- Provides user-friendly error messages with the server URL
Fixes#8568
Related to #8499🤖 Generated with [Letta Code](https://letta.com)
Co-authored-by: letta-code <248085862+letta-code@users.noreply.github.com>
Co-authored-by: Letta <noreply@letta.com>
Co-authored-by: Kian Jones <11655409+kianjones9@users.noreply.github.com>
When multiple concurrent transactions try to upsert the same tools,
they can deadlock if they acquire row locks in different orders.
This fix sorts tools by name before the bulk INSERT to ensure all
transactions acquire locks in a consistent order, preventing deadlocks.
Fixes#8666🤖 Generated with [Letta Code](https://letta.com)
Co-authored-by: letta-code <248085862+letta-code@users.noreply.github.com>
Co-authored-by: datadog-official[bot] <datadog-official[bot]@users.noreply.github.com>
Co-authored-by: Letta <noreply@letta.com>
Co-authored-by: Kian Jones <11655409+kianjones9@users.noreply.github.com>
Add ToolError to exception handling alongside McpError in MCP client classes.
ToolError is raised by fastmcp for input validation errors (e.g., missing
required properties like 'filename'). Both error types are expected user-facing
errors from external MCP servers and should be logged at warning/debug level
to avoid triggering production alerts.
Fixes issue with production error: "fastmcp.exceptions.ToolError: Input
validation error: 'filename' is a required property"
🤖 Generated with [Letta Code](https://letta.com)
Co-authored-by: letta-code <248085862+letta-code@users.noreply.github.com>
Co-authored-by: Letta <noreply@letta.com>
Co-authored-by: datadog-official[bot] <datadog-official[bot]@users.noreply.github.com>
Co-authored-by: Kian Jones <11655409+kianjones9@users.noreply.github.com>
The MCP library internally uses TaskGroup for async operations, which can
raise ExceptionGroup when cleanup fails. This was causing unhandled errors
to propagate in production.
Changes:
- Update cleanup() method in AsyncBaseMCPClient to catch ExceptionGroup
using except* syntax and log errors at debug level (best-effort cleanup)
- Remove redundant try/except blocks in mcp_manager.py and
mcp_server_manager.py that incorrectly re-raised cleanup exceptions
Fixes#8560🐾 Generated with [Letta Code](https://letta.com)
Co-authored-by: letta-code <248085862+letta-code@users.noreply.github.com>
Co-authored-by: Letta <noreply@letta.com>
Co-authored-by: Kian Jones <11655409+kianjones9@users.noreply.github.com>
- Fix typo "upate" -> "update" in TODO comments (mcp_manager.py, mcp_server_manager.py)
- Improve comments in OAuth callback handler to explain why MCPOAuthSession
is used directly (callback is unauthenticated, manager requires actor)
- Clean up variable naming in callback handler
🐾 Generated with [Letta Code](https://letta.com)
Co-authored-by: Letta <noreply@letta.com>
* feat: add tags support to blocks
* fix: add timestamps and org scoping to blocks_tags
Addresses PR feedback:
1. Migration: Added timestamps (created_at, updated_at), soft delete
(is_deleted), audit fields (_created_by_id, _last_updated_by_id),
and organization_id to blocks_tags table for filtering support.
Follows SQLite baseline pattern (composite PK of block_id+tag, no
separate id column) to avoid insert failures.
2. ORM: Relationship already correct with lazy="raise" to prevent
implicit joins and passive_deletes=True for efficient CASCADE deletes.
3. Schema: Changed normalize_tags() from Any to dict for type safety.
4. SQLite: Added blocks_tags to SQLite baseline schema to prevent
table-not-found errors.
5. Code: Updated all tag row inserts to include organization_id.
🐾 Generated with [Letta Code](https://letta.com)
Co-Authored-By: Letta <noreply@letta.com>
* fix: add ORM columns and update SQLite baseline for blocks_tags
Fixes test failures (CompileError: Unconsumed column names: organization_id):
1. ORM: Added organization_id, timestamps, audit fields to BlocksTags
ORM model to match database schema from migrations.
2. SQLite baseline: Added full column set to blocks_tags (organization_id,
timestamps, audit fields) to match PostgreSQL schema.
3. Test: Added 'tags' to expected Block schema fields.
This ensures SQLite and PostgreSQL have matching schemas and the ORM
can consume all columns that the code inserts.
🐾 Generated with [Letta Code](https://letta.com)
Co-Authored-By: Letta <noreply@letta.com>
* revert change to existing alembic migration
* fix: remove passive_deletes and SQLite support for blocks_tags
1. Removed passive_deletes=True from Block.tags relationship to match
AgentsTags pattern (neither have ondelete CASCADE in DB schema).
2. Removed SQLite branch from _replace_block_pivot_rows_async since
blocks_tags table is PostgreSQL-only (migration skips SQLite).
🐾 Generated with [Letta Code](https://letta.com)
Co-Authored-By: Letta <noreply@letta.com>
* api sync
---------
Co-authored-by: Letta <noreply@letta.com>
MarkItDown.convert() does blocking file I/O and CPU-intensive PDF parsing.
This was blocking the event loop during file uploads.
Now wraps the entire markitdown pipeline (tempfile write, convert, cleanup)
in asyncio.to_thread() to run in thread pool.
🐾 Generated with [Letta Code](https://letta.com)
Co-authored-by: Letta <noreply@letta.com>
Critical fixes:
- llm_client_base.send_llm_request() now calls await self.request_async() instead of self.request()
- Remove unused sync get_openai_embedding() that used sync OpenAI client
- Remove deprecated compile_in_thread_async() from Memory
These were blocking the event loop during LLM requests and embeddings.
🐾 Generated with [Letta Code](https://letta.com)
Co-authored-by: Letta <noreply@letta.com>
MCP errors from external servers (e.g., "The specified key does not exist")
are user-facing issues, not system errors. Downgrading the log level from
warning to debug prevents these expected failures from triggering production
alerts in Datadog/Sentry.
Fixes#8370🤖 Generated with [Letta Code](https://letta.com)
Co-authored-by: letta-code <248085862+letta-code@users.noreply.github.com>
Co-authored-by: datadog-official[bot] <datadog-official[bot]@users.noreply.github.com>
Co-authored-by: Kian Jones <11655409+kianjones9@users.noreply.github.com>
* feat: Add conversation_id filtering to message list and search endpoints
Add optional conversation_id parameter to filter messages by conversation:
- client.agents.messages.list
- client.messages.list
- client.messages.search
Changes:
- Added conversation_id field to MessageSearchRequest and SearchAllMessagesRequest schemas
- Added conversation_id filtering to list_messages in message_manager.py
- Updated get_agent_recall_async and get_all_messages_recall_async in server.py
- Added conversation_id query parameter to router endpoints
- Updated Turbopuffer client to support conversation_id filtering in searches
Fixes#8320🤖 Generated with [Letta Code](https://letta.com)
Co-Authored-By: Charles Packer <cpacker@users.noreply.github.com>
* add conversation_id to message and tpuf
* default messages filter for backward compatibility
* add test and auto gen
* fix integration test
* fix test
* update test
---------
Co-authored-by: letta-code <248085862+letta-code@users.noreply.github.com>
Co-authored-by: Charles Packer <cpacker@users.noreply.github.com>
Co-authored-by: christinatong01 <christina@letta.com>
