langgraph

LangGraph is a low-level orchestration framework for building, managing, and deploying long-running, stateful agents. It is developed by LangChain Inc. and can be used standalone or alongside other LangChain products. The project positions itself as infrastructure for any long-running, stateful workflow or agent. Its documented capabilities include durable execution, which lets agents persist through failures and resume from where they left off; human-in-the-loop support, allowing inspection and modification of agent state during execution; comprehensive memory with short-term working memory and long-term persistent memory across sessions; debugging and observability through LangSmith, including tracing of execution paths and state transitions; and production-ready deployment infrastructure for stateful, long-running workflows. Installation is via pip with the langgraph package. A JavaScript/TypeScript equivalent, LangGraph.js, is available separately. The README also points to Deep Agents, a higher-level package built on LangGraph for agents that can plan, use subagents, and leverage file systems. The ecosystem section describes integrations with LangChain for composable components and LLM application development, LangSmith for agent evaluation and observability, and LangSmith Deployment for scaling agents, including visual prototyping in LangSmith Studio. Documentation, API references, a quickstart, guides, a free LangChain Academy course, case studies, a contributing guide, and a code of conduct are linked. The README states that LangGraph is inspired by Pregel and Apache Beam, with a public interface drawing inspiration from NetworkX. It is released under the MIT license. The README mentions adoption by companies including Klarna, Replit, and Elastic, though no independent performance benchmarks or rankings are provided.