langchain-fundamentals

To create a LangChain agent with custom tools, use the create_agent() function to manage the agent loop and state, then define tools using the @tool decorator in Python or tool() in TypeScript and attach them to the agent configuration.

Middleware in LangChain agents works by integrating specialized components like HumanInTheLoopMiddleware to control execution flows, handle errors, and manage optional persistence via MemorySaver during the agent loop.

Yes, you can use create_agent with both Python and TypeScript. The approach provides specific examples for both languages to illustrate common workflows, tool definitions, and quick starts for your development environment.

The best way to add memory persistence to a LangChain agent is by integrating middleware like HumanInTheLoopMiddleware and configuring optional persistence using MemorySaver to retain state across interactions.

You configure prompts and checks for LangChain agent orchestration by defining your requirements within the create_agent() function, which manages the agent loop, configurable prompts, and state validation.

You use middleware for human-in-the-loop control in LangChain to actively manage agent execution flows, intercept errors, and apply necessary checks before proceeding, ensuring safe and controlled agent orchestration.