Model or dataset
eigent-ai/eigent avatar
eigent-ai/eigent

Eigent's quick start talks to its cloud; the local multi-agent desktop takes a separate path

Eigent: The Open Source Cowork Desktop - Local and Free Alternative to Claude Cowork and Codex

15,853 stars1,894 forksTypeScriptApache-2.0

At a glance

What is it?
Eigent is an open-source Electron desktop app built on CAMEL-AI that runs one agent or a coordinated workforce of agents against your files, browser and terminal. The fastest way to launch it connects to Eigent's cloud and needs an account, so anyone choosing it for privacy has to follow the local deployment guide instead.
Who is it for?
Eigent suits people who want a desktop agent workspace they can inspect and modify, with a multi-agent workforce, scheduled runs and their own choice of model, including local inference through vLLM, Ollama or LM Studio. It is not a zero-configuration local tool: the one-command quick start depends on Eigent's cloud and an account, and full local operation means following server/README_EN.md and maintaining both a Node and a Python environment.
Can I use it commercially?
Yes. Apache-2.0 is a permissive licence: you can use, modify and sell software built on it, as long as you keep its copyright and licence notices.
Is it still maintained?
Yes. The repository last received commits 2 days ago.
What is it written in?
Mainly TypeScript, according to GitHub's language statistics.

Answers come from the project's GitHub data, last synced on November 19, 2026, and from our analysis. They are not legal advice.

From daily snapshots

Star history

  • eigent-ai/eigent15,853 ★ · +213
Star history
Repository2026-10-222026-11-19
eigent-ai/eigent15.6k15.9k

213 Stars in the last 28 days, measured from the daily GitHub snapshot. Compare with other projects

Editorial analysis

The quick start connects to Eigent's cloud and needs an account

The shortest route to a running Eigent is four commands, with Node.js 18 to 22 and npm installed:

bash
git clone https://github.com/eigent-ai/eigent.git
cd eigent
npm install
npm run dev

This is the cloud-connected mode. It uses Eigent's cloud backend and requires account registration. That matters because the project's own positioning is a local and free alternative to Claude Cowork and Codex, and a reader who stops at the first code block ends up with neither property: work goes through a hosted backend tied to an account.

The practical rule is to treat npm run dev as a preview. If the reason you picked Eigent is that your files and credentials should stay on your machine, do not evaluate it in this mode and assume the local mode behaves identically. Also note the Node range: versions newer than 22 are outside the stated prerequisite, so a developer machine on a current Node release may need a version manager before npm install.

Local deployment is the recommended mode, documented separately

The mode the project recommends is local deployment: fully standalone, with complete control over your data and no cloud account required. Its instructions live in server/README_EN.md rather than in the main page. The setup includes a local backend server with the full API, local model integration with vLLM, Ollama, LM Studio and similar runtimes, complete isolation from cloud services, and no external dependencies.

This is the configuration behind the Local and Secure claim, where agents run on your machine with local-first execution and your files, credentials and context stay under your control. It is also where the model-agnostic design pays off. Eigent connects to cloud APIs, enterprise gateways or local inference without locking you to one vendor, so a team that already runs Ollama or vLLM can point Eigent at it.

The cost is that the model is your responsibility. Agent quality, speed and how well multi-step plans hold together depend on the model you connect. The advertised Zero Setup applies to the hosted experience; local deployment means running and maintaining the backend server and the inference runtime yourself.

Two dependency stacks: npm for the desktop, uv for the Python backend

Eigent is an Electron and Vite front end in TypeScript with a Python backend in backend/. Each git pull can change either side, and both have to be refreshed:

bash
# 1. Update frontend dependencies (in project root)
npm install

# 2. Update backend/Python dependencies (in backend directory)
cd backend
uv sync

Forgetting the second step leaves the desktop shell on new code while the backend runs with the Python packages from before the pull. The package.json also runs a postinstall script, fix-node-pty-permissions.js, for the node-pty terminal dependency, and the build pipeline chains several repair scripts before compiling: preinstall-deps, compile-babel, fix-venv-paths and fix-symlinks.

Windows takes its own prebuild:win route that skips the clean-symlinks step used on other platforms. Contributors building from source should expect to work in both ecosystems.

One meta agent or a workforce that splits the job

Eigent offers two ways to work. The single-agent harness runs focused tasks with one meta agent, and you research, write, debug and operate alongside it in a desktop workspace. The workforce mode scales to multiple specialized agents that divide the work, collaborate in parallel and carry out multi-step workflows together. This multi-agent workforce comes from CAMEL-AI's open-source project, which Eigent is built on.

Agents get built-in browser and terminal toolkits, and they can be extended through MCP integration and skill integration. An automation feature schedules recurring workflows so agents run tasks at a set time while you are away. One of the featured examples coordinates parallel agents to build ten HTML5 games.

The trade-off of the workforce mode is that parallel agents multiply model calls. With a paid API that means cost, and with local inference it means the hardware has to serve several agents at once. Starting a task in single-agent mode and moving to a workforce only when the task splits cleanly keeps both under control.

Desktop installers come from electron-builder scripts

For people who do not want to build anything, there is a download page at eigent.ai/download. Building your own goes through electron-builder, with separate targets per platform: build:mac, build:win and build:linux, each running the prebuild chain first and publishing nothing. A build:mac:test variant adds a signing test, and the repository carries entitlements.mac.plist for macOS.

There is also a browser-only development mode, dev:web, which runs Vite with vite.config.web.ts instead of the Electron shell, and the debug configuration points the dev server at 127.0.0.1:7777.

The repository also includes a .storybook/ setup, vitest configuration and a test/ directory, along with an AGENTS.md for coding agents working on the codebase. Translated READMEs exist in Portuguese, Simplified Chinese and Japanese.

A license mismatch and features held for the paid tiers

GitHub reports the repository license as Apache-2.0, while package.json declares MIT and marks the package private. Both are permissive, but they are not the same license, and a company redistributing a modified build should confirm with the LICENSE file and the licenses/ directory which terms apply rather than relying on either label.

The open-source app is also not the whole product. The feature list names enterprise features such as SSO, access control and custom enquiries, and the enterprise section describes exclusive features like SSO and custom development, scalable enterprise deployment and negotiated SLAs through a sales contact. A separate cloud version hosts the models, APIs and storage and offers premium support by subscription. A team that needs single sign-on should plan on that conversation with sales, not on a configuration flag in the open-source build.

Release pace is steady: v1.0.5 on 2026-09-25, v1.0.6 on 2026-10-16 and v1.0.7 on 2026-10-30, with the last push to main on 2026-11-02.

Editorial conclusion

Eigent suits people who want a desktop agent workspace they can inspect and modify, with a multi-agent workforce, scheduled runs and their own choice of model, including local inference through vLLM, Ollama or LM Studio. It is not a zero-configuration local tool: the one-command quick start depends on Eigent's cloud and an account, and full local operation means following server/README_EN.md and maintaining both a Node and a Python environment. Before adopting it, confirm which license applies to your use, since GitHub reports Apache-2.0 while package.json says MIT, and check whether features you need, such as SSO, sit in the enterprise tier.

Frequently asked questions

What is Eigent?

Eigent is an open-source desktop app, built on CAMEL-AI, that runs a single agent or a multi-agent workforce on your tasks with browser and terminal toolkits, MCP and skill integration, and scheduled automation.

Is Eigent free?

The Eigent source code is open and can be downloaded, inspected and modified. A hosted cloud version with premium support by subscription and an enterprise tier with features such as SSO are offered separately.

How do I use Eigent?

For a quick preview, clone the repository and run npm install and npm run dev with Node.js 18 to 22, which connects to Eigent's cloud and needs an account. For standalone use, follow the local deployment guide in server/README_EN.md.

Is Eigent AI safe for private data?

In local deployment Eigent runs its own backend isolated from cloud services and can use local models through vLLM, Ollama or LM Studio. The npm run dev quick start instead connects to Eigent cloud services.

Official sources

  1. eigent-ai/eigent on GitHub
  2. License: Apache-2.0
  3. Project website
  4. README
  5. Releases
Add this badge to your README

If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.

Add this badge to your README

markdown
[![Hysen Labs](https://hysenlabs.com/badge/eigent-ai-eigent.svg)](https://hysenlabs.com/projects/eigent-ai-eigent)

Star history chart for your README

markdown
[![Star history](https://hysenlabs.com/chart/eigent-ai-eigent.svg)](https://hysenlabs.com/projects/eigent-ai-eigent)

Was this useful?