pydantic-ai-python

name	pydantic-ai-python
description	Build AI agents with PydanticAI — tools, structured output, multi-provider, streaming, multi-agent delegation, MCP, capabilities, evals, graphs. Use when working with pydantic_ai, pydantic-ai, Agent, RunContext, tool decorator, output_type, deps_type, capabilities, or any PydanticAI agent pattern in Python.

PydanticAI

Python agent framework by the Pydantic team. Model-agnostic, type-safe, dependency-injected.

Installation

uv add pydantic-ai
# Provider extras:
uv add 'pydantic-ai[openai]'    # or anthropic, google, groq, mistral, cohere, bedrock, huggingface
uv add 'pydantic-ai[logfire]'   # observability

Python 3.9+. Requires pydantic v2.

Quick start

from pydantic_ai import Agent

agent = Agent('anthropic:claude-sonnet-4-6', instructions='Be concise.')
result = agent.run_sync('Hello')
print(result.output)

Architecture at a glance

An Agent bundles: instructions, tools/toolsets, output type, dependency type, model, model settings, and capabilities.

Agents are generic: Agent[DepsType, OutputType]. They're designed for reuse — instantiate globally like a FastAPI app.

Running agents

Five modes: run() (async), run_sync(), run_stream() / run_stream_sync(), run_stream_events(), iter() (graph node iteration).

All return result.output typed to your output_type.

Reference files

• agents-and-dependencies.md — Agent construction, instructions, deps, run modes, conversations
• tools-and-toolsets.md — Function tools, toolsets, deferred tools, built-in tools, MCP, common tools
• output-and-streaming.md — Structured output, streaming, validation, output functions
• models-and-providers.md — Provider strings, model settings, fallback, custom models
• capabilities-and-hooks.md — Composable capabilities, hooks, agent specs (YAML/JSON)
• multi-agent.md — Delegation, handoff, programmatic multi-agent patterns
• advanced.md — Multimodal input, thinking, graphs, evals, durable execution, embeddings, extensibility, harness