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
- Monty code as a typed value: tools and hooks written in Monty, in capabilities and agent specs
- Harness: an unattended mode that denies approvals and questions, and reaches sub-agents
- DBOS: run sub-agent tool calls as child workflows, and keep tool calls parallel when a workspace is attached
- Make every harness capability work under durable execution, and enforce it in the test suite
- Call a tool outside an agent run, with validation and capability hooks, and serve a toolset over MCP
- Public read-only access to an agent's tools and capabilities, `Agent.override(capabilities=)`, and exporting `AgentInstructions`
- Let cost and telemetry roll up a conversation across sub-agent runs
- Let code running in a sandbox workspace call the agent's tools through the host
- `ClearToolResults`: target specific tools and count by user turns
- Structured spill for large JSON tool outputs (JSON to a SQL table, shared with sub-agents)
- Docs
- Python not yet supported