Tools
Tools
All ToolsAgent FrameworksVisit Website
LangGraph
MIT28.1K starsFree (LangSmith for monitoring sold separately)MIT
Build stateful, multi-agent applications with cycles and persistence.
About
LangGraph is LangChain's framework for building stateful agent applications. It models agent workflows as graphs — with nodes for actions and edges for transitions — giving you fine-grained control over agent behavior, branching, and state management.
Unlike simpler agent frameworks, LangGraph supports cycles (agents can loop back), conditional edges (dynamic routing), and persistent state (agents can pause and resume). It's the framework of choice for complex, production agent workflows.
Features
- Graph-based agent workflow definition
- Stateful execution with checkpointing
- Cycles and conditional routing
- Human-in-the-loop breakpoints
- Multi-agent coordination primitives
- LangSmith integration for tracing
Use cases
- Complex multi-step agent workflows
- Agents with approval gates and human review
- Multi-agent systems with dynamic delegation
- Long-running workflows that need pause/resume
Alternatives
Claude Agent SDKCrewAIAutoGen
Try LangGraph
Visit the official website to get started.
Building an AI agent tool? Submit it for free or get featured placement for $400.