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
- Leading instructions have no source: MCP server instructions, workspace files and `with_authority` can't be sent below operator voice
- Say in the framework's announcement that the user only speaks in their own messages, so quoted consent in a tool result isn't taken as the user's
- Add a Maritime sandbox capability to Pydantic AI Harness
- `CapabilityCreation` should refuse to load code from a workspace below operator authority
- [aw] Detection Runs
- One JSON transcript renderer for summarizers and judges instead of four tag encodings
- fix: classify media types case-insensitively in is_text_like_media_type
- Parse tags an agent declares out of model text into typed response parts, as `<think>` becomes a `ThinkingPart`
- With `on_change='rewrite'`, Kimi K3.1 Pro, GLM 5.3 Flash and MiniMax drop an instruction change when a tool result contains a lookalike
- `ToolReturn(content=...)` reaches the model as the user's words instead of as part of the tool result
- Docs
- Python not yet supported