PuPu Editor — AI-Powered IDE

An AI-powered development environment built on top of VS Code, similar to Cursor, Windsurf, and Antigravity. Connects to free models hosted on Hugging Face, NVIDIA, Groq, OpenRouter, or your local machine via Ollama.

✨ Features

🤖 Autonomous Agent System

• Plan → Execute → Verify loop — the AI plans tasks, executes them with tools, and verifies results
• Tool-calling — the agent can read/write files, search code, run terminal commands
• Multi-step tasks — handles complex multi-file operations autonomously

🛠️ Built-in Tools

🔌 Model Providers

• Ollama (Local) — Free, private, no API key needed
• Groq — Ultra-fast inference, free tier
• OpenRouter — 300+ models, free models available
• Hugging Face — Free inference API

💬 Chat Panel

• Sidebar chat with streaming responses
• Tool call visualization
• Context-aware (automatically includes active file context)
• Keyboard shortcut: Ctrl+Shift+L

✍️ Inline Completion (Ghost Text)

• Automatic code suggestions as you type
• FIM (Fill-in-the-Middle) support
• Debounced for performance

🔗 MCP Server Support

• Connect to external tools via Model Context Protocol
• Stdio and HTTP transports
• Configure via .pupu/mcp.json

📚 Skills System

• Define reusable instruction sets in .pupu/skills/
• YAML frontmatter + markdown format
• Auto-injected into agent context

📋 Project Rules

• Create .pupurules in your project root
• Rules are automatically included in agent system prompt

🚀 Getting Started

Prerequisites

• VS Code 1.85+
• Node.js 18+
• (Recommended) Ollama installed locally

Install & Run

# Clone the project
cd PuPuEditor

# Install dependencies
npm install

# Build
npm run compile:dev

# Run in VS Code
# Press F5 to open Extension Development Host

Configure a Provider

1. Press Ctrl+Shift+P → "PuPu: Configure Model Providers"
2. Select a provider and enter your API key
3. API keys are stored securely using VS Code's secret storage (OS keychain)

For Ollama (Local, Free)

# Install Ollama
curl -fsSL https://ollama.ai/install.sh | sh

# Pull a model
ollama pull llama3.2

# That's it! PuPu auto-connects to localhost:11434

📁 Project Structure

src/
├── extension.ts                    # Entry point
├── PuPuExtension.ts                # Main extension class
├── types/index.ts                  # All TypeScript interfaces
├── agents/
│   └── AgentOrchestrator.ts        # Agent with tool-calling loop
├── completion/
│   └── InlineCompletionProvider.ts # Ghost text completions
├── config/
│   └── ConfigurationManager.ts     # Settings & secrets
├── managers/
│   ├── ContextManager.ts           # Code context extraction
│   └── ModelRouter.ts              # Provider selection & FIM
├── mcp/
│   └── MCPServerManager.ts         # MCP protocol support
├── providers/
│   ├── OllamaProvider.ts           # Local Ollama
│   ├── OpenAICompatibleProvider.ts # Groq, OpenRouter, HuggingFace
│   └── ProviderRegistry.ts        # Provider lifecycle
├── skills/
│   └── SkillManager.ts             # Skill discovery & loading
├── tools/
│   ├── ToolRegistry.ts             # Tool registration & execution
│   ├── file/                       # File tools
│   ├── search/                     # Search tools
│   └── terminal/                   # Terminal tools
└── ui/
    ├── ChatViewProvider.ts         # Sidebar chat UI
    ├── ChatPanel.ts                # Panel wrapper
    └── StatusBarManager.ts         # Status bar

⌨️ Keyboard Shortcuts

🔧 Configuration

📄 License

MIT