Match sync and async execution
Use the synchronous client for ordinary workers and the async client for asyncio request or agent paths so telemetry does not block the event loop.
Selection framework
Log where the final result is known
Send the event after the transaction finishes
Wait for the database commit, queued task, or model workflow to finish before logging success. A handler can return successfully while that work is still pending.
Categorize exceptions before logging
Record a controlled error type and failed step. Avoid raw exception text when it can contain SQL parameters, prompts, secrets, or customer content.
Shared event contract
Use the same field definitions across services
- operation, status, duration_ms, release, and environment
- request_id, task_id, run_id, account_id, and controlled error_type
- attempt, model, provider, tool_name, and database operation where relevant
Adapt these fields to your workflow. Document units, allowed statuses, and identifiers before several services write to the same table. Name the team responsible for the schema.
Verification
Exercise failure paths before alerts
- 1Exercise one synchronous and one asynchronous path when the application uses both.
- 2Confirm events are emitted after commit or final task status.
- 3Test cancellation, timeout, retry, and process-shutdown behavior.
- 4Inspect inferred field types before building the first query.
Implementation guides
Match the framework, provider, or runtime
Setup guide with source links
Python and FastAPI structured logging
Capture FastAPI request outcomes and latency with the asynchronous Telemetry client and normalized route names.
Open guideSetup guide with source links
Django and Celery structured event monitoring
Connect Django request outcomes and Celery task lifecycles with safe identifiers, latency, retries, and final status.
Open guideSetup guide with source links
Flask and SQLAlchemy telemetry
Record Flask request results and SQLAlchemy transaction behavior with approved Python event fields, consistent route templates, and safe error categories.
Open guideSetup guide with source links
Pydantic AI agent telemetry
Measure Pydantic AI run outcomes, validated outputs, retries, tool activity, latency, usage, and product acceptance with safe structured events.
Open guideSetup guide with source links
CrewAI workflow telemetry
Track CrewAI crew and flow outcomes, task counts, handoffs, retries, latency, cost, and accepted results without retaining agent conversations.
Open guideSetup guide with source links
LlamaIndex agent telemetry
Track LlamaIndex agent and workflow outcomes, tool activity, latency, retrieval versions, cost, and reviewed quality with structured events.
Open guideSetup guide with source links
DSPy program evaluation telemetry
Measure DSPy program latency, failures, metric scores, optimizer versions, model cost, and accepted outcomes across releases with SQL-ready events.
Open guideSetup guide with source links
AutoGen task flow, state, and timeout monitoring
Track AutoGen task-flow state, timeouts, team and agent outcomes, tool activity, handoffs, latency, failures, cost, and releases with compact structured events.
Open guideSetup guide with source links
LiteLLM gateway and SDK telemetry
Track LiteLLM provider routing, retries, fallbacks, token usage, latency, cost estimates, and accepted outcomes across a controlled model gateway.
Open guideVerify one end-to-end workflow first
Create a free API key, choose the closest runtime guide, and send one known success and failure before expanding coverage.