AI agent framework by the Pydantic team. Model-agnostic, type-safe with full observability, human-in-the-loop approvals, durable execution, and integrated MCP/A2A support.
SectorDeveloper Tools InfraNicheAgent Framework Open SourceTypeDeveloper frameworkAgent levelL0 NON Agent NodeAuthorityNoneSourcesai.pydantic.dev · github.com/pydantic/pydantic-aiLast checked2026-05-21
additional metadata
human oversightunknowntask scopeunknownnode scopeframeworkpersistencepersistent identityowner typecommercial ownerregisterabilityclaimable indexed row
We index agent products, platforms, frameworks, APIs, marketplaces, companies, and research demos. L0 means supporting infrastructure. L1–L5 describe increasing agent autonomy. About these classes →
Pydantic-AI is a model-agnostic AI agent framework developed by the Pydantic team. It emphasizes type safety, offers full observability, supports human-in-the-loop approvals, durable execution, and integrates with MCP/A2A protocols.
This is a framework for building AI agents, focusing on type safety and robust execution.
example workflow
Define agent components using Pydantic's type-safe models.
Implement agent logic with LLM integrations.
Configure observability and logging for agent behavior.
snippets: Pydantic AI | Pydantic Docs · Join the world's most widely adopted, AI-powered developer platform where millions of developers, businesses, and the largest open source community build software that advances humanity. · Pydantic AI
