Build agents as controllable step-by-step graphs.
From the LangChain team. You define each step an agent can take and how it moves between them, which makes complex agents easier to debug and run in production. Open source in Python and TypeScript, with optional paid hosting through LangSmith.
Key info
- Pricing
- Open source
- Category
- Agents
- Best for
- Expert
- What it is
- A framework: a code library for building your own agents
- Do you need to code?
- Yes, in Python or TypeScript
- Cost
- Free and open source; optional paid hosting and monitoring through LangSmith
- Models
- Any, through LangChain's integrations
- Install
- pip install langgraph, or npm install @langchain/langgraph
- Good for
- Complex agents that need every step to be controllable in production
Related reading
- Fundamentals
Row level security is the security boundary
If your database rows are protected by a check in your application code, they are not protected. Anything holding a key can talk to the database directly, an…
Expert