// agents · FellouAI/eko

eko

eko ist ein Open-Source-Projekt aus der Kategorie KI-Agenten von FellouAI (FellouAI/eko auf GitHub). Es hat 4.960 Sterne auf GitHub und 445 Forks, wird unter der Lizenz MIT veröffentlicht und wurde zuletzt am 3. März 2026 auf GitHub aktualisiert.

Build production-ready agentic workflows from natural language — agents that operate browsers and computers.

★ 5,0 Tsd. Sterne⑂ 445 ForksMITaktualisiert Mär 2026

Auf GitHub ansehen ↗Official site ↗

▦ Visual — das README dieses Projekts enthält keinen einbettbaren Screenshot,
darum hier ein Platzhalter. Die Anleitung unten ist trotzdem vollständig.

// projektdaten

Projektdaten

Repository: FellouAI/eko
GitHub-Sterne: 4.960
Forks: 445
Lizenz: MIT
Zuletzt auf GitHub aktualisiert: 3. März 2026
Kategorie: KI-Agenten
Eintragstyp: Vollständige Installationsanleitung
Offizielle Website: Official site

// so funktioniert es

So funktioniert es

Clarify the workflow
Describe the multi-step task in natural language — what to open, extract, fill in or produce.
Install the packages
Add the Eko core package (and the nodejs, web or extension package for your environment) with pnpm.
Configure a model key
Provide at least one model API key for the agents to reason with.
Run your first workflow
Start from one of the bundled examples — extension, Node.js automation or web login — and watch the agent execute your description.

// funktionen

Warum es so beliebt ist

Natural-language workflows

Production-ready agentic workflows are built from natural-language descriptions rather than hand-coded step scripts.

Browser and computer operation

Agents perform browser automation and web scraping, system file and process management, and GUI automation.

Dependency-aware parallel agents

Eko 3.0 introduced dependency-aware parallel agent execution, so independent sub-tasks run concurrently in the right order.

Pause, resume and interrupt

Task controls with task_snapshot workflow recovery let long automations be paused, resumed and safely interrupted.

Chat conversations

Eko 4.0 supports chat conversations on top of the workflow engine, with optimized agent logic.

Three deployment shapes

Ships examples and packages for a browser extension, a web application and a Node.js (Playwright) automation host.

// selbst ausführen

Führe es selbst aus

Clone the repository
Clone Eko and enter the monorepo.
```
git clone https://github.com/FellouAI/eko.git
cd eko
```
Install the framework
For your own project, add the core package with pnpm (in the repo, build the workspace packages first).
```
pnpm install @eko-ai/eko
```
Build and run an example
Build the core packages, then run the Node.js/Playwright example with a model API key configured.
```
pnpm install
pnpm build
# Then follow example/nodejs — provide at least one model API key before running
```

// anwendungsfälle

Was du damit bauen kannst

Automating a multi-step web research and data-entry workflow from a plain description
Building a browser-extension agent that fills forms and extracts results
Running parallel agents over independent sub-tasks of a larger automation
Adding pause/resume control to a long-running desktop automation

// häufige fragen

Häufige Fragen

Who maintains Eko?

Eko is maintained by FellouAI (the team behind the Fellou browser), with the product page at fellou.ai/eko.

Is Eko free?

Yes — the framework is open source under the MIT license. You pay for the model APIs your agents call.

Does it need a paid API key?

Yes — the examples require at least one model API key before running, since Eko's agents reason with an LLM you provide.

How does it compare to LangChain or browser-use?

The README publishes its own framework comparison; Eko's emphasis is production workflows described in natural language that operate both the browser and the computer, with parallel agents and pause/resume recovery.

How do I get started?

Add @eko-ai/eko with pnpm, configure a model key, and run the bundled Node.js or extension example.

// weiter entdecken

Alternativen im Bereich KI-Agenten

Die anderen Projekte der Kategorie KI-Agenten mit den meisten Sternen im LibreAI-Katalog — die realistische Auswahl, falls eko nicht das Richtige ist.

openclaw★ 391,6 Tsd.
A personal AI that really does things — on any OS, with your data staying yours.
superpowers★ 294,9 Tsd.
An agentic skills framework & software development methodology that works.
ECC★ 275,2 Tsd.
Agent harness optimization system with skills, memory and security for Claude Code, Codex and more
hermes-agent★ 252,0 Tsd.
The agent that grows with you — a Python framework for adaptive AI agents.
deepseek-harness★ 245,5 Tsd.
DeepSeek's plugin-based agent harness where everything is a plugin.

// mehr in agents

Mehr in KI-Agenten

Verwandte Projekte der Kategorie KI-Agenten im Katalog, die eko bei den Sternen am nächsten stehen — zum Weiterstöbern auf beiden Seiten.

ha-mcp★ 5,0 Tsd.
The unofficial Home Assistant MCP server — control and automate your smart home from any AI assistant.
zep★ 5,0 Tsd.
Examples and integrations for Zep — the memory layer for AI agents.
ag2★ 5,0 Tsd.
The open-source AgentOS, formerly known as AutoGen.
OpenMausBot★ 4,9 Tsd.
OpenMausBot — the open-source Grok Bot alternative: your own team of AI bots in a chat app, each a real local agent with its own personality, model and cloud computer
exa-mcp-server★ 5,1 Tsd.
Exa's MCP server for web search and crawling — give your AI agent a live, high-quality web index.
agent-device★ 4,9 Tsd.
Mobile app automation and verification built for AI coding agents — CLI, MCP server, and Node API

Hol dir eko

Kostenlos und offen — hol dir den Quellcode, gib dem Repository einen Stern und leg los.

Auf GitHub ansehen ↗← Mehr Tools: KI-Agenten

eko