deepagents

Deep Agents is an agent harness. An opinionated, ready-to-run agent out of the box. Instead of wiring up prompts, tools, and context management yourself, you get a working agent immediately and customize what you need.

What's included:

• Planning — write_todos for task breakdown and progress tracking
• Filesystem — read_file, write_file, edit_file, ls, glob, grep for reading and writing context
• Shell access — execute for running commands (with sandboxing)
• Sub-agents — task for delegating work with isolated context windows
• Smart defaults — Prompts that teach the model how to use these tools effectively
• Context management — Auto-summarization when conversations get long, large outputs saved to files

pip install deepagents
# or
uv add deepagents

from deepagents import create_deep_agent

agent = create_deep_agent()
result = agent.invoke({"messages": [{"role": "user", "content": "Research LangGraph and write a summary"}]})

The agent can plan, read/write files, and manage its own context. Add tools, customize prompts, or swap models as needed.

Add your own tools, swap models, customize prompts, configure sub-agents, and more. See the documentation for full details.

from langchain.chat_models import init_chat_model

agent = create_deep_agent(
    model=init_chat_model("openai:gpt-4o"),
    tools=[my_custom_tool],
    system_prompt="You are a research assistant.",
)

MCP is supported via langchain-mcp-adapters.

Try Deep Agents instantly from the terminal:

uv tool install deepagents-cli
deepagents

The CLI adds conversation resume, web search, remote sandboxes (Modal, Runloop, Daytona, & more), persistent memory, custom skills, headless mode, and human-in-the-loop approval. See the CLI documentation for more.

create_deep_agent returns a compiled LangGraph graph. Use it with streaming, Studio, checkpointers, or any LangGraph feature.

• 100% open source — MIT licensed, fully extensible
• Provider agnostic — Works with Claude, OpenAI, Google, or any LangChain-compatible model
• Built on LangGraph — Production-ready runtime with streaming, persistence, and checkpointing
• Batteries included — Planning, file access, sub-agents, and context management work out of the box
• Get started in seconds — pip install deepagents or uv add deepagents and you have a working agent
• Customize in minutes — Add tools, swap models, tune prompts when you need to

• docs.langchain.com – Comprehensive documentation, including conceptual overviews and guides
• reference.langchain.com/python – API reference docs for Deep Agents packages
• Chat LangChain – Chat with the LangChain documentation and get answers to your questions

Discussions: Visit the LangChain Forum to connect with the community and share all of your technical questions, ideas, and feedback.

• Examples — Working agents and patterns
• API Reference – Detailed reference on navigating base packages and integrations for LangChain.
• Contributing Guide – Learn how to contribute to LangChain projects and find good first issues.
• Code of Conduct – Our community guidelines and standards for participation.

This is a monorepo containing all Deep Agents packages:

Package	PyPI	Description
deepagents		Core SDK — create_deep_agent, middleware, backends
deepagents-cli		Interactive terminal interface with TUI, web search, and sandboxes
deepagents-acp		Agent Client Protocol integration for editors like Zed
deepagents-harbor	-	Harbor evaluation and benchmark framework
langchain-daytona		Daytona sandbox integration
langchain-modal		Modal sandbox integration
langchain-runloop		Runloop sandbox integration

This project was primarily inspired by Claude Code, and initially was largely an attempt to see what made Claude Code general purpose, and make it even more so.

Deep Agents follows a "trust the LLM" model. The agent can do anything its tools allow. Enforce boundaries at the tool/sandbox level, not by expecting the model to self-police.