EverFern

Your autonomous AI desktop agent — free, open-source, and runs entirely on your machine.

EverFern is a free, open-source AI desktop agent that automates complex tasks on your computer using natural language. Think of it as your personal AI coworker that can:

Research topics and synthesize findings
Write and edit code, documents, and presentations
Browse the web and interact with websites
Operate your desktop GUI (click, type, navigate)
Process files, analyze data, and generate reports

EverFern is the open-source alternative to Manus Desktop, Claude Cowork, and similar commercial AI agents — available to everyone, completely free.

Why EverFern?

Feature	EverFern	Commercial Alternatives
Price	Free & open-source	$200+/month
Data Privacy	Everything stays local	Cloud processing required
Self-Hosting	Run on your own hardware	Not supported
Customization	Full source access	Proprietary, locked
AI Providers	Local or cloud of your choice	Provider-locked

✨ Features

🤖 Autonomous Task Execution — Tell EverFern what you want in plain English. It plans, reasons, and executes multi-step tasks autonomously.
🖥️ Desktop Automation — EverFern can control your mouse and keyboard, take screenshots, and interact with any application just like a human would.
🔍 Web Research — Browse websites, extract information, compare products, and summarize findings automatically.
📄 Document Processing — Read, analyze, and create documents in formats like PDF, Word, Excel, PowerPoint, CSV, and more.
💻 Code Assistant — Write, review, debug, and refactor code across multiple languages with full project context.
🔌 Extensible Tools — Connect to 30+ built-in tools or add custom ones via MCP (Model Context Protocol).
🛠️ Multi-Provider Support — Use local AI (Ollama, LMStudio) for privacy or cloud providers (OpenAI, Anthropic, DeepSeek, etc.) for power.
🔒 Privacy-First — All data stays on your machine. No cloud dependency. Your conversations never leave your desktop.

🚀 Quick Start

Prerequisites

Node.js v18+
Windows 10/11, macOS, or Linux

Installation

# Clone the repository
git clone https://github.com/everfern/desktop.git
cd desktop

# Install dependencies
npm install

# Start the application
npm run dev

Build for Production

npm run build
npm run make

💬 How It Works

Just tell EverFern what you need:

"Research the top 5 AI coding assistants and create a comparison table"

"Summarize all the PDF reports in my Downloads folder"

"Open VS Code and refactor the auth module to use JWT tokens"

"Create a PowerPoint presentation about our Q1 sales data"

"Find and organize all my photos from last year into albums"

EverFern breaks down your request, plans the steps, and executes them — showing you its thinking process in real-time.

🏗️ Architecture

┌─────────────────────────────────────────────────────────────┐
│                    Your Desktop                             │
├─────────────────────────────────────────────────────────────┤
│  ┌─────────────────────────────────────────────────────┐    │
│  │              EverFern (Frontend + Backend)           │    │
│  │  ┌──────────────────┐    ┌────────────────────────┐  │    │
│  │  │   React UI       │    │   Agent Engine          │  │    │
│  │  │   (Chat + Tools) │◄──►│   (LangGraph-based)    │  │    │
│  │  └──────────────────┘    └────────────────────────┘  │    │
│  └─────────────────────────┬────────────────────────────┘    │
│                            │                                 │
│         ┌──────────────────┼──────────────────┐              │
│         │                  │                  │              │
│    ┌────▼────┐       ┌─────▼─────┐      ┌─────▼─────┐       │
│    │Local AI │       │Cloud AI   │      │  Tools    │       │
│    │Ollama   │       │Anthropic  │      │  GUI Auto │       │
│    │LMStudio │       │OpenAI     │      │  Web Scrape│       │
│    └─────────┘       └───────────┘      └───────────┘       │
└─────────────────────────────────────────────────────────────┘

Frontend: Next.js with real-time streaming UI
Backend: Electron with LangGraph-powered agent orchestration
AI: Your choice of local or cloud providers

🔧 Configuration

AI Provider Setup

EverFern supports multiple AI providers. Configure yours in the app settings:

Local (Privacy-First)

Ollama — Run models locally
LMStudio — Local model server

Cloud (More Power)

OpenAI (GPT-4, GPT-4o)
Anthropic (Claude 3.5, Opus)
DeepSeek
Google Gemini
And more...

Environment Variables

NEXT_TELEMETRY_DISABLED=1    # Disable telemetry
UV_THREADPOOL_SIZE=4         # Async operation threads

📂 Project Structure

everfern/
├── src/                    # Next.js frontend
│   ├── app/                # App router (chat, settings)
│   └── components/         # React components
├── main/                   # Electron backend
│   ├── agent/              # Agent orchestration
│   │   ├── runner/         # Graph execution nodes
│   │   ├── tools/          # Built-in tools
│   │   └── skills/         # Document processing
│   └── acp/                # AI provider clients
├── docs/                   # Architecture docs
└── public/                 # Static assets

🔐 Privacy & Security

Local Storage — All data, keys, and history stored in ~/.everfern/store
No Cloud Sync — Your conversations never leave your machine
Your Keys — API keys encrypted and stored locally
Open Source — Audit the code yourself for trust

🤝 Contributing

Contributions welcome! Whether it's bug reports, features, or code:

Issues — Report bugs on GitHub
Discussions — Join community conversations
Code — Submit pull requests

📄 License

MIT License — free for personal and commercial use.

🙏 Built With

LangGraph — Agent orchestration
Next.js — Frontend framework
Electron — Desktop application framework
TypeScript — Type-safe development

Made with ❤️ by the EverFern Community

Everfern