Open Computer Use

AI agents that control computers like humans do.

Browser automation · Terminal access · Desktop control · Multi-agent orchestration

Website · Discord · Twitter

Marketing — Market your product on Reddit autonomously View chat session	Go-to-Market — Find prospects and send personalized emails View chat session
QA Testing — Test every checkout flow and report bugs View chat session	Job Application — Find roles, tailor your resume, and apply View chat session
Form Filling — Fill out the YC S26 application for you View chat session	Social Media — Post on Hacker News and engage with comments View chat session

Open Computer Use is an open-source platform that gives AI agents real computer control. Unlike chatbots that only talk about tasks, agents here actually perform them — browsing the web, running commands, clicking through UIs, and orchestrating multi-step workflows in isolated containers.

Computer use capabilities similar to Anthropic's Claude Computer Use, but fully open-source and extensible.

Browser — Search-first web navigation, form filling, element interaction, multi-tab management, screenshot capture.

Terminal — Command execution, file operations, script running, package management, output streaming.

Desktop — Mouse & keyboard control, window management, screenshot analysis, UI element detection via computer vision.

Planner — Decomposes complex requests into subtasks, assigns to specialized agents, passes context between steps.

Frontend (Next.js 15)         Backend (FastAPI)              VM (Docker)
┌──────────────────┐     ┌─────────────────────────┐     ┌──────────────────┐
│  Chat UI         │────▶│  Multi-Agent Executor    │────▶│  Chrome Browser  │
│  Model Selector  │ SSE │  ├─ Planner Agent        │ WS  │  Terminal        │
│  VM Management   │◀────│  ├─ Browser Agent        │◀────│  Desktop (XFCE)  │
│  Zustand Stores  │     │  ├─ Terminal Agent       │     │  Agent Server    │
└──────────────────┘     │  └─ Desktop Agent        │     │  VNC :5900       │
                         │  WebSocket · DB · Billing│     └──────────────────┘
                         └─────────────────────────┘

You only need one API key to run the app. Get a free sandbox key at coasty.ai/developers — sandbox keys (sk-coasty-test-*) return mock infrastructure for free; live keys provision real VMs.

git clone https://github.com/coasty-ai/open-computer-use.git
cd open-computer-use
npm install
cp .env.oss.example .env.local

Open .env.local and paste your key:

COASTY_API_KEY=sk-coasty-test-your-key-here

Then:

npm run dev

Open http://localhost:3000 and start a chat. That's it — no Supabase, no AWS, no Stripe to configure.

First boot auto-generates CSRF_SECRET and ENCRYPTION_KEY into .env.local (already gitignored). Safe to delete; we'll regenerate.

cd backend
python -m venv venv
source venv/bin/activate    # Windows: venv\Scripts\activate
pip install -r requirements.txt
cd ..

cp .env.example .env.local
cp backend/.env.example backend/.env

# Supabase
NEXT_PUBLIC_SUPABASE_URL=https://your-project.supabase.co
NEXT_PUBLIC_SUPABASE_ANON_KEY=your-anon-key
SUPABASE_SERVICE_ROLE=your-service-role-key

# Security — generate each with: openssl rand -hex 32 (or base64 32 for ENCRYPTION_KEY)
ENCRYPTION_KEY=
CSRF_SECRET=
INTERNAL_API_KEY=          # MUST match in both files

# AWS Bedrock (LLM)
AWS_ACCESS_KEY_ID=
AWS_SECRET_ACCESS_KEY=
AWS_REGION=us-east-1
BEDROCK_DEFAULT_MODEL=anthropic.claude-sonnet-4-20250514-v1:0

# Optional: web-search tool
GOOGLE_SEARCH_KEY=
GOOGLE_SEARCH_CX=

npm install -g supabase
supabase login
supabase link --project-ref your-project-ref
supabase db push

# Terminal 1 — Frontend
npm run dev

# Terminal 2 — Backend (dev)
cd backend && python main.py

# Terminal 3 — Optional: local AI desktop VM
docker-compose -f docker-compose.ai-desktop.yml up --build

docker-compose up --build

npx @coasty/mcp

A lightweight overlay that runs AI agent commands directly on your local machine instead of a remote VM.

• Floating always-on-top pill UI with expanded chat panel
• Platform-native automation (PowerShell/Win32 on Windows, CoreGraphics/osascript on macOS, xdotool on Linux)
• Browser control via Puppeteer-core, shell execution, file operations
• WebSocket bridge to backend with auto-reconnect

cd electron
npm install
npm run dev        # Development with hot reload
npm run package    # Build for current platform

├── app/                    # Next.js routes & pages
├── components/             # React components (UI, chat, prompts)
├── lib/                    # Stores, providers, services, utilities
├── backend/
│   └── app/
│       ├── api/routes/     # FastAPI endpoints
│       ├── services/       # Multi-agent executor, VM control, billing
│       ├── providers/      # AI provider integrations
│       └── core/           # Config, middleware, logging
├── electron/
│   └── src/
│       ├── main/           # App lifecycle, IPC, automation modules
│       ├── preload/        # Context bridge API
│       └── renderer/       # React UI, stores, components
├── docker/ai-desktop/      # Ubuntu VM container
└── supabase/               # Database schema

1. Fork the repo
2. Create a branch: git checkout -b feature/your-feature
3. Commit your changes
4. Open a pull request

Bug reports and feature requests welcome in Issues.

• Multi-VM parallel orchestration
• Visual workflow builder
• Agent marketplace & templates
• Windows / macOS VM support
• Plugin system for custom tools
• Collaborative sessions
• Voice control & video understanding

This platform gives AI agents significant autonomy. Use it to automate repetitive tasks, testing, research, and content creation — not to violate terms of service, spam, or scrape without permission. Always use isolated environments, respect robots.txt, and follow data protection laws.

Apache License 2.0 — Copyright (c) 2025 Open Computer Use Contributors