eko is a free, open source ai development platforms project written in TypeScript and released under MIT. It has 4,961 GitHub stars, 443 forks and 15 open issues, and was last pushed 7 months ago. On this registry it ranks #85 of 116 tracked projects in AI Development Platforms, with 5 head-to-head comparisons available.

What is eko?

Eko (Eko Keeps Operating, pronounced like "echo") is a production-ready, MIT-licensed JavaScript and TypeScript framework for building reliable AI agents — from a single natural-language command up to complex multi-step workflows — behind one unified interface for both computer and browser environments, aimed at developers who want agentic automation inside the JavaScript ecosystem rather than a server-only stack.

What it is

Eko is a pure JavaScript agentic framework distributed as a pnpm-managed monorepo, with packages for Node.js, the browser, and a browser extension (@eko-ai/eko, @eko-ai/eko-nodejs, @eko-ai/eko-web, @eko-ai/eko-extension). A developer supplies an LLMs map and a list of agents, then hands both to a single Eko instance that plans and executes the work. Version 4.0 added chat conversations and revised agent logic; version 3.0 introduced dependency-aware parallel agent execution along with pause, resume, and interrupt controls backed by task_snapshot workflow recovery.

The problem it solves is fragmentation. Agent work that spans a browser and a local machine has usually meant pairing a server-side framework with a separate browser automation library and writing the glue between them. Eko replaces that split with one interface: the same agents run in computer and browser environments, and multi-agent coordination, planning, and interruption live in one place. Because the framework is JavaScript-first, a product team does not have to leave its existing stack to ship an agentic workflow.

Key capabilities

  • Pure JavaScript core: runs in browsers and Node.js, so agents ship alongside existing front-end and back-end code.
  • Multi-agent tasks: several agents cooperate in one workflow; the quickstart instantiates new BrowserAgent() and new FileAgent() and passes them to new Eko({ llms, agents }).
  • Native MCP support: connects to the Awesome MCP Servers catalogue as a tool source.
  • Dynamic LLM routing: one LLMs map mixes anthropic (claude-sonnet-4-5-20250929), google (gemini-2.5-pro), openai (gpt-5), and OpenAI-compatible endpoints such as qwen-plus and doubao-seed-1-6-250615 configured through baseURL.
  • Human-in-the-loop control: pause, resume, and interrupt running work, with task_snapshot for workflow recovery.
  • Dependency-aware parallel execution: independent agent steps run concurrently rather than strictly in sequence.
  • Stream planning plus loop and listener tasks: dynamic rendering of plans and automation of repetitive jobs.

Who uses it and how

  • Product teams building conversational assistants, which Eko 4.0 targets with chat conversations and optimized agent logic.
  • Teams whose workflows reach private web resources the open web cannot, combining browser agents with file and machine access.
  • Front-end and extension developers, where @eko-ai/eko-web and @eko-ai/eko-extension cover the browser runtimes and @eko-ai/eko-nodejs carries the server side.
  • Operations that need human review gates, using pause, resume, interrupt, and task_snapshot to let a person approve or correct a step mid-run.

Getting started

Install or upgrade through pnpm with pnpm up @eko-ai/eko @eko-ai/eko-nodejs @eko-ai/eko-web @eko-ai/eko-extension, or clean-reinstall with rm -rf node_modules && pnpm install. Full setup and configuration instructions live in the Eko Quickstart guide at https://eko.fellou.ai/docs/getting-started/quickstart/.

How it compares

The framework's own comparison table names Langchain, Browser-use, Dify.ai and Coze. Eko records itself as supporting all platforms, with one-sentence-to-multi-step workflows, intervenability, parallel tasks, and access to private web resources, where Langchain is server-side, Browser-use is browser-only, Dify.ai is web-hosted, and Coze is web-hosted and not open-source. Eko, Langchain, Browser-use and Dify.ai are all open-source, so the licence axis does not separate them; platform coverage, parallelism, and private-resource access do.

When to use it — and when not to

Two listed capabilities, Observable Chain and Native A2A, are marked as coming soon and are not part of the shipped feature set. The README is deliberately short and defers setup, security, and configuration to the external documentation site, so teams wanting everything in-repo should expect to read docs. Browser and front-end deployments must not embed API keys — credentials belong behind a backend proxy configured through baseURL and request headers — and the 4.0 upgrade regenerates saved workflows to a v3 schema, so existing exported plans need rework before migrating.

project readme (upstream, from github) — read inline

eko-logo
Eko - Build Production-ready Agentic Workflow with Natural Language

License Build Status Version

Eko (pronounced like 'echo') is a production-ready JavaScript framework that enables developers to create reliable agents, from simple commands to complex workflows. It provides a unified interface for running agents in both computer and browser environments.

News

  • 2025-11: Eko 4.0 supports chat conversations & optimizes agent logic.
  • 2025-09: Eko 3.0 introduces dependency-aware parallel agent execution.
  • 2025-09: New pause, resume, and interrupt controls with task_snapshot workflow recovery.
  • 2025-09: Monorepo tooling migrated to pnpm for consistent workspace management.

Upgrading to Eko 4.0

Follow these steps when moving an existing Eko 3.x project to 4.0:

  1. Update dependencies with pnpm up @eko-ai/eko @eko-ai/eko-nodejs @eko-ai/eko-web @eko-ai/eko-extension.
  2. Regenerate saved workflows or exported plans so they use the v3 schema and dependency graph format.
  3. Clean and reinstall using pnpm (rm -rf node_modules && pnpm install), then rebuild any browser or desktop bundles.
  4. Re-run automated demos and update documentation to reflect the new pause/interrupt APIs and parallel agent behavior.

Framework Comparison

Feature Eko Langchain Browser-use Dify.ai Coze
Supported Platform All platform Server side Browser Web Web
One sentence to multi-step workflow
Intervenability
Task Parallel
Development Efficiency High Low Middle Middle Low
Task Complexity High High Low Middle Middle
Open-source
Access to private web resources

Features

  • Pure Built for browsers and Node.js.🚀
  • Multi-Agent: Unleash power with multiple Agents in one task.📈
  • Agent/Tool Flexibility: Customize new Agents and Tools in just one line.🎉
  • Native MCP: Connects seamlessly with Awesome MCP Servers.🔗
  • Dynamic LLM: Balance speed and performance with flexible model choices.⚙️
  • Human-in-the-loop: Intervene when it matters most.🤝
  • Stream Planning: Dynamic rendering made easy.🎨
  • Loop & Listener Tasks: Automate any repetitive task.🤖
  • Observable Chain: Coming soon
  • Native A2A: Coming soon

Quickstart

Note: Please refer to the Eko Quickstart guide guide for full instructions on how to run it.

Security Warning

DO NOT use API Keys in browser/frontend code!

This will expose your credentials and may lead to unauthorized usage.

Best Practices: Configure backend API proxy request through baseURL and request headers.

Please refer to the link: https://eko.fellou.ai/docs/getting-started/configuration#web-environment

const llms: LLMs = {
  default: {
    provider: "anthropic",
    model: "claude-sonnet-4-5-20250929",
    apiKey: "your-api-key"
  },
  gemini: {
    provider: "google",
    model: "gemini-2.5-pro",
    apiKey: "your-api-key"
  },
  openai: {
    provider: "openai",
    model: "gpt-5",
    apiKey: "your-api-key"
  },
  // OpenAI-compatible models (Qwen, Doubao, etc.)
  qwen: {
    provider: "openai",
    model: "qwen-plus",
    apiKey: "your-qwen-api-key",
    config: {
      baseURL: "https://dashscope-intl.aliyuncs.com/compatible-mode/v1"
    }
  },
  doubao: {
    provider: "openai",  // Use OpenAI provider for compatibility
    model: "doubao-seed-1-6-250615",  // or other Doubao model
    apiKey: "your-volcengine-api-key",
    config: {
      baseURL: "https://ark.cn-beijing.volces.com/api/v3"  // Volcengine endpoint
    }
  }

};

let agents: Agent[] = [new BrowserAgent(), new FileAgent()];
let eko = new Eko({ llms, agents });
let result = await eko.run("Search for the latest news about Musk, summarize and save to the desktop as Musk.md");
$ pnpm install @eko-ai/eko

Example Projects

The repository ships with three workspace examples under the example/ folder.

Prerequisites

Before running any example, install dependencies and build the core packages from the root directory:

pnpm install
pnpm build

Browser Extension (example/extension)

cd example/extension
pnpm install
pnpm run build

Load the generated dist directory via chrome://extensions → Developer Mode → Load unpacked. Configure your API key in the extension options before running the automation task.

Node.js Automation (example/nodejs)

cd example/nodejs
pnpm install
pnpm playwright install   # first time only, installs browsers
pnpm run build
OPENAI_API_KEY=... ANTHROPIC_API_KEY=... pnpm run start

The Node.js demo drives Playwright through Eko; provide at least one model API key before running it.

Web Login Demo (example/web)

cd example/web
pnpm install
pnpm run start

This starts a React dev server on the default port with a simple login flow that you can automate with the browser or web agents.

Use Cases

  • Browser automation and web scraping
  • System file and process management
  • Workflow automation
  • Data processing and organization
  • GUI automation
  • Multi-step task orchestration

Documentation

Visit our documentation site for:

  • Getting started guide
  • API reference
  • Usage examples
  • Best practices
  • Configuration options

Development Environments

Eko can be used in multiple environments:

  • Browser Extension
  • Web Applications
  • Node.js Applications

Community and Support

Star History Chart

License

Eko is released under the MIT License. See the LICENSE file for details.

Frequently asked questions

Is eko free to use?

eko is open source under the MIT licence. There is no licence fee and no seat count — you can self-host it or, where the project offers one, pay a vendor for a managed version instead.

What does eko do?

Eko (Eko Keeps Operating) - Build Production-ready Agentic Workflow with Natural Language - eko.fellou.ai

What is eko written in?

eko is primarily written in TypeScript. Its source is publicly available at https://github.com/FellouAI/eko, and it has 4,961 GitHub stars.