Python Integration Reference

Web Frameworks

HTTP Clients

Databases

AI/LLM Frameworks

PydanticAI, OpenAI, and Anthropic need no Logfire extra — install the library itself (uv add pydantic-ai / uv add openai / uv add anthropic); Logfire's own AI extras are litellm, dspy, and google-genai.

import logfire

logfire.configure()
logfire.instrument_pydantic_ai()  # captures agent runs, tool calls, LLM request/response
# or:
logfire.instrument_openai()       # captures chat completions, embeddings, token counts
logfire.instrument_anthropic()    # captures messages, token usage

For PydanticAI, each agent run becomes a parent span containing child spans for every tool call and LLM request. See the main skill's Agent Frameworks table for coverage depth across other frameworks (LangChain, CrewAI, AutoGen, ...).

Task Queues

Other

gateway, datasets, and variables are extras too, but for separate product features, not app instrumentation: the AI Gateway proxy, the evals SDK (see the logfire-evals skill), and managed feature flags respectively.

Gunicorn Configuration

# gunicorn.conf.py
import logfire


def post_fork(server, worker):
    logfire.configure()


def post_worker_init(worker):
    logfire.instrument_flask(worker.wsgi)

This example is specific to Flask: configure Logfire in post_fork, then instrument the loaded Flask application in post_worker_init. This preserves Gunicorn's application-loading lifecycle when preload mode is disabled and installs instrumentation in each worker when it is enabled. Other frameworks have different call signatures and placement requirements; follow their row in the table above and their integration documentation rather than mechanically substituting another instrument_*() call here.