What problem does it solve? Building production agents with LangChain requires knowing the current recommended APIs and avoiding common pitfalls like missing persistence, runaway loops, and vague tool descriptions. This Skill provides the canonical patterns for creating agents with create_agent, defining tools, adding middleware, and handling errors in both Python and TypeScript. ## Core Features & Use Cases - Agent Creation with create_agent: Configure models, tools, system prompts, checkpointers, and middleware using the recommended API. - Middleware Patterns: Implement human-in-the-loop approval workflows and custom tool-call hooks with HumanInTheLoopMiddleware and wrap_tool_call. - Common Fixes: Resolve missing tool descriptions, absent checkpointers, infinite loops via recursion_limit, and incorrect result access. - Use Case: You need a support agent that remembers conversation context, requires human approval before executing dangerous tools, and returns typed structured responses. This Skill gives you the exact code patterns for all three requirements. ## Quick Start Create a LangChain agent with create_agent that uses a weather tool, persists state with MemorySaver, and requires human approval for dangerous actions.
