• Picks
  • Discover
  • Alternatives
  • About

© 2026 Willet

  • Newsletter
  • About
  • FAQ
  • Terms
  • Privacy & Cookies
Join the Discord (opens in a new tab)
LangChain avatar

Deep Agents

Opinionated Python agent harness with sub-agents, files, and memory

  • Agent frameworks and tools
  • AI coding agents
View on GitHub
Website

What it is

Deep Agents is a Python library built on LangGraph that gives you a ready-to-run agent with planning, a pluggable filesystem, shell access in a sandbox, sub-agents with separate context windows, context summarization, persistent memory, and human approval of tool calls. Developers building long-running, multi-step agents start from its defaults and then swap models, tools, prompts, or any component without forking. It works with any tool-calling model, including local ones via Ollama or vLLM, and accepts MCP servers as tools. A companion terminal coding agent, Deep Agents Code, is built on it.

Images

Facts

Stars
30,450 GitHub stars.
30,450

Forks
4,293
Commits
4,280
Contributors
162
Language
Python
License
MIT
Last commit
Oct 20, 2026
On GitHub since
July 2025
First release
November 2025
Platforms
CLI, library
Self-hostable
No
Tags
  • multi-agent
  • agent-framework
  • agent-harness
  • ai-memory
  • context-engineering
  • ai-agents

Last synced from GitHub on Oct 20, 2026

The 10 best overlooked open-source projects, every Monday.

Share and more

Comments

No comments yet. Start the conversation.

Similar projects

  • AgentScope-AI avatar

    AgentScope

    Python framework for building multi-agent and realtime voice LLM agents

    33.5k
  • strands-agents avatar

    Strands Agents

    Open-source Python and TypeScript toolkit for writing AI agents

    8.9k
  • Astro avatar

    Flue

    TypeScript framework for building sandboxed autonomous agents

    8.5k