LangChain just open-sourced a Claude Code alternative
DeepAgents gives you a working AI coding agent out of the box. Planning, file editing, and sub-agents. Open-source, MIT licensed.
LangChain, the company behind one of the most popular AI development frameworks, just dropped DeepAgents — an open-source agent framework that works like a ready-to-go AI coding assistant. Think of it as a Claude Code alternative that you can customize, extend, and run with any AI model you want.
The project hit 14,800+ GitHub stars with over 1,200 new stars just today, making it one of the hottest open-source projects right now.
An AI Coding Agent That Works Out of the Box
Most AI agent frameworks require you to wire up prompts, tools, and context management from scratch. DeepAgents flips that: you get a working agent immediately, then customize what you need.
Here's what's built in:
Planning — The agent breaks complex tasks into steps and tracks its own progress, like a built-in to-do list for AI
File operations — It can read, write, edit, and search through your project files
Shell commands — Runs terminal commands with optional sandboxing (isolated safe environment) so it can't break anything
Sub-agents — Can spawn smaller, focused helper agents for specific parts of a task, each with their own workspace
Works With Any AI Model — Not Just Claude
Unlike Claude Code (which only uses Anthropic's models), DeepAgents is model-agnostic. You can plug in OpenAI's GPT, Google's Gemini, open-source models running locally via Ollama, or any model that supports tool calling. Switch models with a single line of code.
The project also supports MCP (Model Context Protocol) — the same standard that Claude Code uses to connect to external tools — so you can add databases, APIs, and custom integrations.
Who Should Try This
Developers who want full control — If you love what Claude Code does but want to customize everything (the model, the prompts, the tools), DeepAgents gives you that freedom.
Teams building internal tools — Since it's MIT licensed (completely free, even for commercial use), companies can build custom AI agents for their specific workflows without paying per-seat fees.
AI tinkerers — The CLI version includes web search, persistent memory, and human-in-the-loop approval (where you review actions before the AI takes them).
Try It Now
Getting started takes two lines:
pip install deepagentsThen in Python:
from deepagents import create_deep_agent
agent = create_deep_agent()
result = agent.invoke({"messages": [{"role": "user", "content": "Research LangGraph and write a summary"}]})Or install the CLI for a terminal-based experience:
curl -LsSf https://raw.githubusercontent.com/langchain-ai/deepagents/main/libs/cli/scripts/install.sh | bashThe Bigger Picture
The creators explicitly say DeepAgents was inspired by Claude Code — they studied what made it effective and built an open-source version that anyone can extend. With LangChain's massive developer ecosystem behind it, this could become the default starting point for teams building custom AI coding agents.
The race to democratize AI coding tools is accelerating. Claude Code set the standard; now open-source alternatives are catching up fast.
Check out the DeepAgents GitHub repository or read the documentation to get started.
Related Content — Get Started with Easy Claude Code | Free Learning Guides | More AI News
Stay updated on AI news
Simple explanations of the latest AI developments