Devtoolwatch BoardDeep Agents review
Deep Agents is an open source, opinionated agent harness from LangChain, built on LangGraph, with sub-agents, filesystem, context management, and skills bundled in.
TL;DR
Deep Agents is an open source agent harness that runs out of the box and lets you extend, override, or replace any piece. It suits developers building long-horizon, multi-step agents in Python or TypeScript. It bundles sub-agents, a pluggable filesystem, context management, memory, and human-in-the-loop on top of LangGraph, and works with any tool-calling model.
- 30.8kGitHub stars
- +294new stars, week of Oct 25
- ~2.5Mdownloads a week, +13% vs 4-week avg
- #9in Agent frameworks this week
Deep Agents at a glance
- Category
- Agent frameworks
- Maintained by
- LangChain (organization, @langchain-ai on GitHub)
- License
- MIT
- Language
- Python, Shell, JavaScript
- Repository
- github.com/langchain-ai/deepagents
- Website
- docs.langchain.com/deepagents
- npm package
- deepagents 1.15.1
- PyPI package
- deepagents 0.7.29
- Latest release
- deepagents-code==0.1.87 on GitHub, Oct 20, 2026
- First published
- Jul 27, 2025
What Deep Agents does
- Deep Agents delegates tasks to sub-agents that run with isolated context windows.
- Deep Agents reads, writes, edits, and searches files over pluggable local, sandboxed, or remote backends.
- Deep Agents summarizes long threads and offloads tool outputs to disk to manage growing context.
- Deep Agents runs shell commands in a sandbox of your choice.
- Deep Agents provides persistent memory through pluggable state and store backends for cross-session recall.
- Deep Agents lets a human approve, edit, or reject tool calls before they run.
- Deep Agents accepts your own functions or any MCP server as tools, plus skills the agent loads on demand.
How it works
You install the deepagents Python package and call create_deep_agent with a model string, a system prompt, and optional tools, then invoke the agent with messages. The library sits on LangChain's create_agent and the LangGraph runtime, which provides streaming, persistence, and checkpointing. Tracing, evaluation, and deployment connect through LangSmith. A JavaScript/TypeScript version is published as deepagents on npm.
How to install and start using Deep Agents
With uv
uv add deepagentsDeep Agents Code (terminal coding agent)
curl -LsSf https://langch.in/dcode | bashRequirements
- Python 3.11 or later, below 4.0.
- An LLM that supports tool calling, from a hosted API, an open-weight provider, or a self-hosted runtime.
First steps
- Add the package to your project with uv add deepagents.
- Import create_deep_agent from deepagents and pass a model string and a system prompt.
- Optionally pass your own Python functions as tools when creating the agent.
- Call agent.invoke with a messages payload to run the agent.
Who Deep Agents is for
- Developers who want planning, context management, and delegation out of the box instead of building them.
- Teams building long-horizon, multi-step agents for production on LangGraph and LangSmith.
- Projects that need to switch between frontier, open-weight, or local models.
Limitations to know
- Deep Agents follows a trust-the-LLM model: the agent can do anything its tools allow, so you must enforce boundaries at the tool or sandbox level.
- Deep Agents works only with models that support tool calling.
- Deep Agents is opinionated; for a lighter harness without bundled middleware, LangChain suggests its create_agent instead.
- Deep Agents uses an agent loop; agents needing a different shape require a custom LangGraph graph.
Is Deep Agents growing? Usage data
In the week of Oct 25–31, 2026, Deep Agents ranked #92 on the Devtoolwatch Board and #9 in Agent frameworks, with a momentum score of 65 out of 100. Its main package had about 2.5M downloads that week, +13% against its own 4-week average.
Show the weekly numbers
| Week of | New stars |
|---|---|
| Aug 9, 2026 | 285 |
| Aug 16, 2026 | 299 |
| Aug 23, 2026 | 638 |
| Aug 30, 2026 | 345 |
| Sep 6, 2026 | 323 |
| Sep 13, 2026 | 247 |
| Sep 20, 2026 | 228 |
| Sep 27, 2026 | 149 |
| Oct 4, 2026 | 198 |
| Oct 11, 2026 | 278 |
| Oct 18, 2026 | 205 |
| Oct 25, 2026 | 294 |
On Hacker News
- Deep Agent Code Capabilities, 5 points and 0 comments (Jul 3, 2026)
The Board ranks tools by four signals each week: new stars, new stars for the repo's size, added downloads and download growth beyond the market. How the ranking works.
Deep Agents alternatives
Other AI agent frameworks on the Board in the week of Oct 25–31, 2026, by momentum:
| Rank | Tool | Stars | Downloads/wk |
|---|---|---|---|
| #1 | Harness²Harness² is a Python framework from Google Research for recursive agent harnessing, refining an agent's prompts, tools, memory and… | 21.7k | – |
| #2 | HyperFramesHyperFrames is an open-source framework that turns HTML, CSS, media, and seekable animations into deterministic MP4 videos, built for AI… | 76.6k | 1.5M |
| #3 | AgentScopeAgentScope is a Python agent framework with an SDK for building ReAct agents and an agent service backend with a web UI. | 35k | 51.4k |
| #4 | AG-UIAG-UI is an open, event-based protocol that standardizes how AI agents connect to user-facing frontend applications. | 16.9k | 4.6M |
| #5 | LangGraphLangGraph is a low-level orchestration framework and runtime for building, managing, and deploying long-running, stateful agents in… | 44.2k | 18.5M |
| #6 | BifrostOpen-source, high-performance AI gateway that works as your LLM gateway, MCP gateway and agent gateway. One OpenAI-compatible API for… | 9.3k | – |
Recent Deep Agents releases
- deepagents-code==0.1.87, Oct 20, 2026
- deepagents==0.7.29, Oct 20, 2026
- deepagents-code==0.1.86, Oct 16, 2026
- deepagents==0.7.28, Oct 16, 2026
- deepagents==0.7.27, Oct 16, 2026
Deep Agents FAQ
Is Deep Agents free and open source?
Yes. Deep Agents's source code is on GitHub (langchain-ai/deepagents) under the MIT license, so you can use it for free.
How do I install Deep Agents?
Run uv add deepagents. The install section above has the other options from the project's README.
Is Deep Agents still maintained?
Yes. The langchain-ai/deepagents repository was last updated on Oct 21, 2026, and the latest release, deepagents-code==0.1.87 on GitHub, came out on Oct 20, 2026.
How popular is Deep Agents?
Deep Agents has 30.8k GitHub stars as of Nov 1, 2026. It gained 294 new stars in the week of Oct 25–31, 2026, against a 12-week average of 291 a week. Its main package had about 2.5M downloads that week, +13% against its own 4-week average. That made it #9 among AI agent frameworks on the Devtoolwatch Board.
Deep Agents vs LangGraph: how do they compare?
Deep Agents and LangGraph are both tracked by Devtoolwatch as AI agent frameworks. Deep Agents has 30.8k GitHub stars and LangGraph has 44.2k. In the week of Oct 25–31, 2026, Deep Agents's main package had 2.5M downloads and LangGraph's had 18.5M. LangGraph is a low-level orchestration framework and runtime for building, managing, and deploying long-running, stateful agents in Python or JavaScript. LangGraph review.
Deep Agents vs Claude Agent SDK: how do they compare?
Deep Agents and Claude Agent SDK are both tracked by Devtoolwatch as AI agent frameworks. Deep Agents has 30.8k GitHub stars and Claude Agent SDK has 8.4k. In the week of Oct 25–31, 2026, Deep Agents's main package had 2.5M downloads and Claude Agent SDK's had 25.6M. Claude Agent SDK is a Python SDK for building agents that interact programmatically with Claude Code, including its tools, custom tools, and hooks. Claude Agent SDK review.
What are the best alternatives to Deep Agents?
Other AI agent frameworks on the Devtoolwatch Board in the week of Oct 25–31, 2026, by momentum: Harness² (#1), HyperFrames (#2), AgentScope (#3), AG-UI (#4), LangGraph (#5). The alternatives table above has their numbers.
What is Deep Agents?
Deep Agents is an open source agent harness from LangChain: an opinionated agent that runs out of the box. It bundles sub-agents, a filesystem, context management, shell access, persistent memory, human-in-the-loop, skills, and tools, and lets you override or replace any piece without forking.
How does Deep Agents relate to LangChain and LangGraph?
Deep Agents is a layer in the same stack. LangGraph is the graph runtime, LangChain's create_agent is a minimal harness on top of it, and Deep Agents is a more opinionated harness on top of create_agent. Any LangGraph CompiledStateGraph can be passed in as a sub-agent.
Does Deep Agents work with open-weight or local models?
Yes. Deep Agents works with any model that supports tool calling, including OpenAI, Anthropic, and Google APIs, open-weight models on Baseten or Fireworks, and self-hosted models via Ollama, vLLM, or llama.cpp.
What is Deep Agents Code?
Deep Agents Code is a pre-built coding agent that runs in your terminal and can be powered by any LLM. The README compares it to Claude Code or Cursor. Its releases add features such as resuming threads in non-interactive mode and sandbox providers like NVIDIA OpenShell.
Can Deep Agents be used in production?
Yes. Deep Agents is built on LangGraph, which provides streaming, persistence, and checkpointing, and is designed for production agent deployments. LangChain recommends pairing it with LangSmith for tracing, evaluation, and monitoring.
Is Deep Agents available for JavaScript or TypeScript?
Yes. Deep Agents is available as a JavaScript/TypeScript library, deepagents.js, published on npm as deepagents.
Sources
- langchain-ai/deepagents on GitHub
- Deep Agents README
- deepagents on npm
- deepagents on PyPI
- Deep Agents website
- Release deepagents-code==0.1.87 (Oct 20, 2026)
- Release deepagents==0.7.29 (Oct 20, 2026)
- Release deepagents-code==0.1.86 (Oct 16, 2026)
- Hacker News: Deep Agent Code Capabilities
- Devtoolwatch Board, the week of Oct 25–31, 2026
Deep Agents README badge
Maintainers can show the tool's weekly Board rank in their README:
[](https://index.devtoolwatch.com/tools/deepagents/)How this page is made: Devtoolwatch writes each review from the project's README, releases, package metadata and website, with an AI model, then checks every claim against the source it cites and drops what the sources don't support. The numbers come from Devtoolwatch's daily GitHub snapshots, GitHub's stargazer history, and the npm and PyPI download APIs. Spotted an error? Email [email protected].