deepagents — Batteries-Included LangGraph Agent Harness

deepagents is an opinionated, production-ready AI agent framework built on LangGraph. Inspired by Claude Code, it provides immediate agent capabilities without requiring manual assembly of prompts, tools, and context management. Returns a compiled LangGraph graph — works natively with streaming, LangGraph Studio, and checkpointers.

When to use this skill

• Building a Python AI agent that needs file access, shell execution, and todo management out of the box
• Adding memory, skills, or sub-agent orchestration via composable middleware
• Integrating HITL (Human-in-the-Loop) approval workflows before destructive operations
• Replacing manual LangGraph StateGraph assembly with a pre-wired harness
• Using any tool-calling model (Claude, GPT-4o, Gemini) with a unified provider:model interface
• Deploying agents to LangGraph Cloud / LangGraph Studio

Installation