RAG Frameworks · FellouAI

eko

Eko is a TypeScript framework for building production-ready AI agents and multi-step workflows that run in browsers, Node.js, and browser extensions. It supports natural language task descriptions, parallel agent execution, and human-in-the-loop controls, with integrations for multiple LLM providers.

Talk to DEV.co Discuss Implementation

Source: GitHub — github.com/FellouAI/eko

4.9k

GitHub stars

440

Forks

TypeScript

Primary language

MIT

License (OSI-approved)

Key facts

Objective fields from the source. Values we can't verify are shown as “Unknown” rather than guessed.

Field	Value
Repository	FellouAI/eko
Owner	FellouAI
Primary language	TypeScript
License	MIT — OSI-approved
Stars	4.9k
Forks	440
Open issues	11
Latest release	v4.1.0 (2025-12-29)
Last updated	2026-03-03
Source	https://github.com/FellouAI/eko

What eko is

Eko provides a unified agentic framework with support for multiple agents (BrowserAgent, FileAgent, etc.), dynamic LLM selection (Anthropic, OpenAI, Google, OpenAI-compatible providers), MCP server integration, task snapshotting for recovery, and dependency-aware parallel execution. Built in TypeScript with pnpm workspace management and examples for Node.js, browser extension, and React web environments.

Quickstart

Get the eko source

Clone the repository and explore it locally.

terminalbash

1git clone https://github.com/FellouAI/eko.git2cd eko3# follow the project's README for install & configuration

Need it deployed, integrated, or customized instead? DEV.co ships production installs.

Best use cases

Browser and GUI Automation

Automate web scraping, form filling, and multi-step browser workflows via BrowserAgent. Natural language task descriptions simplify complex automation logic compared to low-level Selenium/Playwright code.

Multi-Agent Workflow Orchestration

Coordinate multiple specialized agents (file system, browser, custom tools) in parallel with dependency awareness. Pause, resume, and interrupt workflows mid-execution with task snapshots for resilience.

Human-in-the-Loop Process Automation

Build workflows that pause for human intervention, approval, or clarification. Useful for approval chains, data validation, and scenarios requiring judgment before proceeding.

Implementation considerations

Secure LLM API keys via backend proxy in browser/web environments; never expose credentials in frontend code (documented security warning).
Requires pnpm for workspace management; review monorepo migration from npm if upgrading from older versions.
Version 4.0 introduced breaking changes (pause/resume APIs, parallel agent behavior); plan migration from v3 with workflow regeneration and schema updates.
Multi-agent execution requires careful dependency graph design to avoid deadlocks or unexpected parallelization.
MCP server integration requires compatible MCP implementations; test with your specific server ecosystem.

When to avoid it — and what to weigh

Strict Server-Only Deployment Required — Eko's strength is cross-platform (browser, Node.js, extension) support. If you need pure server-side agents only, Langchain may be more focused.
Simple Sequential LLM Chains — Eko is optimized for agentic workflows with agent autonomy and tool use. For basic prompt-chaining or RAG without agents, lighter frameworks may suffice.
Closed-Source or Proprietary License Requirement — Eko is MIT-licensed open-source. If your policy requires proprietary tooling, this is not a fit.
No JavaScript/TypeScript Stack — Eko is JavaScript/TypeScript only. Python-first or polyglot teams may prefer Langchain or other language-agnostic frameworks.

License & commercial use

MIT License—permissive, allows commercial use, modification, and distribution with attribution. No patent clauses or liability disclaimers beyond standard MIT terms.

MIT permits commercial use without restrictions or royalty obligations. Verify compliance with LLM provider terms (Anthropic, OpenAI, Google) separately, as their API usage policies apply independently. No additional licensing fees from Eko itself.

DEV.co evaluation signals

Editorial assessment — not user reviews. Directional, with an explicit confidence level.

Signal	Assessment
Maintenance	Active
Documentation	Adequate
License clarity	Clear
Deployment complexity	Moderate
DEV.co fit	Strong
Assessment confidence	High

Security considerations

Documentation explicitly warns against exposing API keys in browser/frontend code; recommends backend proxy via baseURL and custom headers. No public security audit, CVE disclosures, or formal threat model documented. Agents execute LLM-generated instructions (e.g., file operations, browser navigation); validate/sandbox task inputs if untrusted. MCP integration security depends on server implementations used.

Alternatives to consider

Langchain (JS/Python)

Server-side agent framework with broader tool ecosystem and community; lacks native browser/extension support and cross-platform unification. Better for pure backend agentic workflows.

Browser-use

Browser-only agent framework focused on GUI automation; simpler but less suitable for multi-environment or complex multi-agent orchestration. No intervenability or parallel execution.

Anthropic Builds (Anthropic SDK)

Lightweight, model-specific agents via Claude's native tool-use; tightly coupled to Anthropic; minimal multi-agent or workflow orchestration features.

Software development agency

Build on eko with DEV.co software developers

Eko is ideal for teams needing cross-platform agent automation with natural language task descriptions. Start with the quickstart guide, review security best practices for API key handling, and explore example projects in the monorepo.

Talk to DEV.co

Related open-source tools

Surfaced by semantic similarity across the DEV.co open-source index.

esp-ai ragbits eliza

Related on DEV.co

Explore the category and the services that help you build with it.

RAG Frameworks AI Application Development Web App Development API Development Talk to DEV.co

eko FAQ

Can I run Eko agents in a production SaaS application?

Yes, but use a backend proxy to secure LLM API keys. Frontend-only deployments expose credentials; Node.js backend examples show the correct pattern. Browser extension deployment requires distribution (Chrome Web Store, etc.).

What happens if an agent fails mid-workflow?

Eko v3.0+ supports task snapshots and pause/resume/interrupt APIs. Workflows can recover from snapshots, but explicit error handling and retry logic depend on your task configuration.

Does Eko work offline?

No. Agents rely on LLM API calls (Anthropic, OpenAI, etc.) which require internet connectivity. Local LLM models via OpenAI-compatible endpoints (e.g., Ollama) may work if your endpoint is reachable.

Is Eko suitable for real-time user interactions?

Yes, especially with human-in-the-loop controls. Stream Planning and Intervenability features allow pausing for user feedback. Latency depends on LLM provider response time.

Software developers & web developers for hire

From first prototype to production, DEV.co delivers software development services around tools like eko. Our software development agency staffs experienced software developers and web developers for custom software development, web development, integrations, and ongoing support across rag frameworks and beyond.

Ready to Build Agentic Workflows?

Talk to DEV.co Schedule a Consultation