Integrations / Pydantic AI
Trace Pydantic AI agents
To trace Pydantic AI with Uncoil, wrap the agent run in trace_agent and record tools as steps, or send Pydantic AI's OpenTelemetry spans to Uncoil over OTLP.
Your code
Pydantic AI
Uncoil
One run.
- 01One trace per agent run
- 02Tool calls and validation failures
- 03Model usage and cost
- 04Retries in order
Setup
A few lines.
python
import uncoil
@uncoil.trace_agent(name="pydantic-agent")
async def ask(question: str):
with uncoil.span("agent-run", type="agent", input=question):
result = await agent.run(question)
return result.outputFull details in the documentation. All integrations are listed on the integrations page.
- Will I see structured-output retries?
- Yes. Each retry is a separate step inside the run, so you can see why validation failed.
