pydantic-ai
https://github.com/pydantic/pydantic-ai
Python
GenAI Agent Framework, the Pydantic way
Triage Issues!
When you volunteer to triage issues, you'll receive an email each day with a link to an open issue that needs help in this project. You'll also receive instructions on how to triage issues.
Triage Docs!
Receive a documented method or class from your favorite GitHub repos in your inbox every day. If you're really pro, receive undocumented methods or classes and supercharge your commit history.
Python not yet supported6 Subscribers
Add a CodeTriage badge to pydantic-ai
Help out
- Issues
- Let an evaluator answer several measures in one model request
- Split `UserError` so a refusal a better model could answer can reach a `FallbackModel`
- No cassette covers a Jev run where a filled tool call and a union meet, and `state` changes shape mid-run
- Describe a field by its docstring consistently, wherever the type is used
- A `functools.partial` passed to `ToolOutput` is treated as a type, adding a parameter that should not exist
- An output function sees its own unanswered output tool call in `ctx.messages`, but only under tool output
- fix(bedrock): omit, don't rewrite, an unsigned trailing thinking block on Anthropic
- List Harness `ToolCallJudge` in capability documentation
- Add a `StructuredJudge` evaluator that answers several measures in one model request
- A `Model` cannot tell messages from `message_history` apart from the ones the agent just built
- Docs
- Python not yet supported