Search for a command to run...
Agent frameworks are the libraries you reach for when you stop using a coding tool and start building one. They give you the agent loop, tool calling, handoffs, and guardrails so you don't reimplement them per project. This section covers the major frameworks and the orchestration patterns that apply across all of them.
Not sure which to pick? The Framework Comparison lays them out side by side. If you're designing a multi-step or multi-agent system, read Agent Orchestration Patterns first — the choice of framework matters less than the shape of the system.
Claude Agent SDK, OpenAI Agents SDK, Pydantic AI, LangChain, CrewAI, and more — when to use each.
Common patterns for building multi-step and multi-agent AI systems.
Claude Code as a library — the same tools, agent loop, and context management, in Python or TypeScript.
Lightweight, OpenAI-native framework with minimal abstractions — Agents, Handoffs, Guardrails.
Model-agnostic Python framework from the Pydantic team — tight validation and Logfire observability.