An intelligent, autonomous AI agent that generates full-stack production-ready projects from a single prompt.
๐ค Type a prompt. Watch the agent work. Get a complete project โ instantly.
DevForge is an AI-powered software factory built on the gitagent framework. It takes a natural language prompt (e.g., "create a chatbot app using React") and autonomously scaffolds a fully structured, ready-to-run project โ complete with source code, configuration, dependencies, and documentation.
It ships with a real-time React dashboard where you can:
- Submit prompts and watch live terminal output stream in
- See all generated files in a dynamic file explorer
- Preview the final generated project structure
| Feature | Description |
|---|---|
| ๐ง AI Pipeline Orchestration | Multi-step pipeline that thinks, designs, codes, tests, and documents |
| โก Real-time Terminal Streaming | Live output logs streamed to the UI via Server-Sent Events (SSE) |
| ๐ Dynamic File Explorer | Real generated file tree shown in the UI after each run |
| ๐ Smart Project Detection | Detects project type from prompt keywords โ React, Node API, etc. |
| ๐งช Test Generation Skill | Auto-generates test scaffolding for generated projects |
| ๐ Debug Skill | Autonomous project debugging pipeline |
| ๐ README Auto-generation | Creates professional documentation per project |
| ๐ฏ Multi-skill Agent | Modular skill system โ add new capabilities without touching core logic |
devforge-agent/
โ
โโโ agent.yaml # Agent configuration (skills, version, metadata)
โโโ SOUL.md # Agent identity & personality
โโโ RULES.md # Agent operating constraints
โ
โโโ skills/ # ๐ง Modular AI skills
โ โโโ generate-react-app/ # React app scaffolding skill
โ โโโ generate-node-api/ # Node.js REST API skill
โ โโโ debug-project/ # Auto-debugging skill
โ โโโ test-generator/ # Test suite generation skill
โ โโโ pipeline-orchestrator/ # Master skill that chains others
โ โโโ create-readme/ # Documentation generation skill
โ โโโ deploy-project/ # Deployment automation skill
โ โโโ project-explainer/ # Code explanation skill
โ โโโ analyze-project/ # Project analysis skill
โ
โโโ dashboard/ # ๐ฅ React + Vite real-time UI
โ โโโ src/
โ โ โโโ pages/Dashboard.jsx # Main dashboard page
โ โ โโโ components/
โ โ โ โโโ Terminal.jsx # Live terminal with SSE streaming
โ โ โ โโโ FileExplorer.jsx# Dynamic real file tree viewer
โ โ โ โโโ PromptBox.jsx # Prompt input with loading state
โ โ โ โโโ Navbar.jsx # Navigation bar
โ โ โ โโโ PreviewPanel.jsx# Output preview panel
โ โโโ server.js # Express API bridge (port 3001)
โ
โโโ demo-output/ # ๐ฆ All AI-generated projects land here
โโโ demo-prompts/ # Sample prompts for quick testing
โโโ mock-claude.js # Smart project generator (agent engine)
โโโ gitclaw / gitclaw.cmd # CLI runner scripts
โโโ run-demo.bat # One-click demo launcher (Windows)
When you submit a prompt, DevForge runs a 7-step autonomous pipeline:
Step 1 โ ๐ Analyze project requirements from prompt
Step 2 โ ๐ Design system architecture & folder structure
Step 3 โ ๐ฆ Create project manifest (package.json, configs)
Step 4 โ ๐ Generate core application source code
Step 5 โ ๐งช Scaffold automated test suites
Step 6 โ ๐ Generate professional documentation (README.md)
Step 7 โ โ
Validate all files & report success
All steps stream to the UI terminal in real time. Generated files appear in the File Explorer automatically when the pipeline completes.
| Prompt Keywords | Generated Project |
|---|---|
react, chatbot, ui, frontend |
React App with App.jsx, components/, index.css, index.html |
node, api, express, todo, rest |
Node.js REST API with routes, middleware, .env.example |
next, nextjs |
Next.js App Router scaffold (coming soon) |
ecommerce, shop |
Full-stack E-commerce scaffold (coming soon) |
- Node.js
>= 18.x - npm
>= 9.x
git clone https://github.com/VishnuVardhanCodes/devforge-ai-agent.git
cd devforge-ai-agentnpm installcd dashboard
npm installnpm run devThis runs both the Vite (React) frontend and the Express API bridge concurrently:
- ๐ Frontend:
http://localhost:5173 - ๐ API Server:
http://localhost:3001
- Open
http://localhost:5173in your browser - Type a prompt in the input box (e.g., "create a chatbot application using React")
- Click Generate Project
- Watch the terminal stream live logs
- See your generated files appear in the File Explorer!
You can also run DevForge from the command line:
Windows:
run-demo.bat demo-prompts\react-app.txtAny OS:
node mock-claude.js -p "create a todo API using express"The generated project will be saved in demo-output/.
Try these prompts in the dashboard or CLI:
"create a chatbot application using react"
"build a todo REST API using express and node.js"
"generate an e-commerce backend API"
"create a react frontend with authentication"
DevForge is built on a modular skills architecture. Each skill is a self-contained unit with its own SKILL.md prompt and configuration.
To add a new skill:
- Create a folder under
skills/your-skill-name/ - Write a
SKILL.mdwith the prompt and metadata - Register it in
agent.yamlunderskills:
skills:
- generate-react-app
- generate-node-api
- debug-project
- your-new-skill # โ just add it hereโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
โ โก DevForge AI Software Factory v0.2.0 โ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโฌโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโค
โ ๐ Generated Files โ โ โ โ gitclaw output โ
โ demo-output/ โ โ
โ โโโ react-chatbot/ โ DevForge v0.2.0 initialized โ
โ โโโ ๐ package.json โ ๐ Received prompt... โ
โ โโโ ๐ README.md โ โถ Invoking pipeline... โ
โ โโโ ๐ index.html โ โถ [1/7] generate-project โ
โ โโโ ๐ src/ โ โ Created: package.json โ
โ โโโ ๐ก App.jsx โ โ Created: App.jsx โ
โ โโโ ๐ก index.jsx โ โ Created: ChatWindow.jsx โ
โ โโโ ๐ components/ โ โ
Pipeline completed! โ
โ โโโ ๐ก ChatWindow.jsx โ โ
โ โโโ ๐ก MessageInput.jsx โ _ โ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโดโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโค
โ ๐ข Live Preview / API Status localhost:3000 โ
โโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโโ
Contributions, issues and feature requests are welcome!
- Fork the repository
- Create your feature branch:
git checkout -b feature/amazing-skill - Commit your changes:
git commit -m 'Add amazing skill' - Push to the branch:
git push origin feature/amazing-skill - Open a Pull Request
This project is licensed under the MIT License โ see the LICENSE file for details.