One command to plan, code, test, debug, and deploy entire applications.
Getting Started · Commands · Agents · Models · Deployment · API
DevOS is an AI-powered CLI tool that acts as your full development team. Describe what you want to build in plain English, and DevOS will plan, code, test, debug, and deploy it — automatically.
npx devos-ai goal "Build a Twitter clone with auth and deploy to Vercel"DevOS will:
- 📋 Plan — Break down your goal into tasks
- 💻 Code — Generate the entire codebase
▶️ Run — Start the application- 🔧 Debug — Catch and fix errors automatically
- 📝 Review — Check code quality
- 🚀 Deploy — Ship to production
npm install -g devos-ai# Set up AI (pick one)
devos config set OPENAI_API_KEY=sk-your-key
# OR use free local models:
ollama pull llama3.3 && devos model add ollama
# Create a project
devos init "Express REST API with JWT auth and PostgreSQL"
# Run with auto-debugging
devos run
# Or go full autopilot
devos goal "Build and deploy a blog with React, markdown support, and Vercel"| Feature | Description |
|---|---|
| 🎯 Goal Mode | Describe what you want → get a deployed app |
| 🔄 Auto-Debug Loop | Runs your app, catches errors, fixes them automatically |
| 🤖 Multi-Agent System | 7 specialized AI agents collaborate on tasks |
| 🧠 Smart Router | Auto-selects the best AI model per task |
| 🌐 40+ AI Models | OpenAI GPT-5.4, Claude 4.7, Gemini 3.1, Ollama, OpenRouter |
| 🖥️ Terminal GUI | Interactive dashboard — just run devos |
| 📁 Workspace Folder | Set where AI creates files with devos config set workspace=/path |
| 🚀 One-Click Deploy | Vercel, Railway, Fly.io, Docker |
| 🔍 Code Search (RAG) | Natural language search across your codebase |
| 🛡️ Security Scanner | Detects secrets, SQL injection, XSS, eval() |
| 💰 Cost Tracking | Real-time AI usage costs per provider |
| 📊 Health Monitor | Auto-healing deployment monitoring |
| 🔀 Framework Migration | AI-powered migration between frameworks |
| ⌨️ Interactive Shell | Natural language command interface |
User Input → Orchestrator → Planner → Coder → Executor → Debugger → Reviewer → Deployer
↑______________Auto-Fix Loop______________↓
| Agent | Role |
|---|---|
| 🎯 Orchestrator | Coordinates all agents, routes tasks |
| 📋 Planner | Breaks goals into structured task plans |
| 💻 Coder | Generates, edits, and refactors code |
| 🔧 Debugger | Analyzes errors and auto-generates fixes |
| 📝 Reviewer | Code quality review and security checks |
| Runs processes and tests | |
| 🚀 Deployer | Handles multi-platform deployment |
| 📊 Monitor | Health checks with self-healing |
| Provider | Models | Setup |
|---|---|---|
| OpenAI | GPT-5.4 Pro, GPT-5 Turbo, o3, o4-mini | devos config set OPENAI_API_KEY=sk-... |
| Anthropic | Claude 4.7 Sonnet/Opus, Claude 4.6 Opus/Sonnet | devos config set ANTHROPIC_API_KEY=sk-ant-... |
| Gemini 3.1 Pro/Flash, 2.5 Pro | devos config set GOOGLE_API_KEY=AIza... |
|
| OpenRouter | 40+ models via single API | devos config set OPENROUTER_API_KEY=sk-or-... |
| Provider | Models | Setup |
|---|---|---|
| Ollama | Llama 3.3, CodeLlama, DeepSeek, Mistral, Qwen | ollama pull llama3.3 && devos model add ollama |
devos model auto # Automatically picks the best model per task- Simple tasks → cheap/fast model (GPT-4o-mini, Gemini 3.1 Flash)
- Medium tasks → balanced model (GPT-5.4 Pro, Claude 4.7 Sonnet)
- Complex tasks → premium model (Claude 4.7 Opus, Gemini 3.1 Pro)
Saves 60-80% on AI costs.
# Project
devos init [description] # Create new project
devos run [script] # Run with auto-fix
devos goal <description> # Full autopilot
devos auto # Auto-debug loop
# Development
devos edit <file> <instruction> # AI-powered editing
devos test [file] # Generate & run tests
devos refactor [glob] # AI refactoring
devos explain [file|error] # Explain code/errors
devos find <query> # RAG-powered code search
devos watch # File watcher
devos shell # Interactive AI shell
# AI & Config
devos model list|use|auto|add # Manage models
devos config set|list|reset # Configuration
devos cost # AI usage costs
# Deployment
devos deploy [--target=platform]# Deploy anywhere
devos rollback [version] # Rollback deployment
devos env set|list|sync # Environment vars
devos monitor [url] # Health monitoring
devos status # Project status
devos logs [service] # View logs
# Tools
devos security # Security scan
devos migrate <target> # Framework migration# Auto-detect platform
devos deploy
# Or specify target
devos deploy --target=vercel
devos deploy --target=railway
devos deploy --target=fly
devos deploy --target=dockerDevOS auto-detects your project type:
- Next.js / React → Vercel
- Express / API → Railway
- Full-stack → Fly.io
- Any project → Docker
import { Orchestrator, AIAdapter } from 'devos-ai';
const orchestrator = new Orchestrator();
await orchestrator.executeGoal('Build a REST API with Express');
// Or use individual agents
import { CoderAgent, DeployerAgent } from 'devos-ai';
const coder = new CoderAgent();
await coder.generateFile('server.js', 'Express server with CORS and rate limiting');devos-ai/
├── bin/devos.js # CLI entry point
├── src/
│ ├── cli/
│ │ ├── index.js # Command registration
│ │ └── commands/ # 22 CLI commands
│ ├── agents/
│ │ ├── orchestrator.js # Agent coordinator
│ │ ├── planner.js # Task planning
│ │ ├── coder.js # Code generation
│ │ ├── debugger.js # Error analysis
│ │ ├── reviewer.js # Code review
│ │ ├── executor.js # Process management
│ │ ├── deployer.js # Deployment
│ │ └── monitor.js # Health monitoring
│ ├── ai/
│ │ ├── adapter.js # Unified AI interface
│ │ ├── router.js # Smart model routing
│ │ └── providers/ # OpenAI, Anthropic, Google, Ollama, OpenRouter
│ ├── code/analyzer.js # Static analysis
│ ├── context/
│ │ ├── rag.js # Code search engine
│ │ └── memory.js # Project memory
│ └── utils/ # Config, logger, security, cost, git
├── docs/ # HTML documentation website
├── package.json
├── LICENSE
└── README.md
- Node.js ≥ 18.0.0
- AI Provider: At least one — OpenAI, Anthropic, Google, Ollama (free), or OpenRouter
- Deployment (optional): Vercel CLI, Railway CLI, Fly CLI, or Docker
MIT © 2025 DevOS Team