Ragrabbit

🚀 RagRabbit - Intelligent Document Retrieval and Generation System

RagRabbit is an intelligent document retrieval and generation platform based on large language models. It supports quick location of document content through semantic search and provides context - related analysis functions.

✨ Features

• Intelligent Retrieval Engine: Utilizes vector databases to achieve semantic similarity matching.
• Context - Related Analysis: Supports multi - round conversations based on text mining technology.
• Knowledge Graph Construction: Automatically generates associations between documents.
• Pluggable AI Models: Supports multiple large language models (e.g., Llama series).

📦 Installation

Basic Installation

Install Dependencies

pnpm install @ragrabbit/core @ragrabbit/ui

Initialize the Configuration File

Create a configuration file named config.json:

{
  "apiPort": 3000,
  "vectorDBConfig": {
    "type": "chroma",
    "host": "localhost",
    "port": 2000
  },
  "llmModel": {
    "type": "llama2"
  }
}

Start the Service

pnpm run dev

Configuration Guide

Intelligent Retrieval Module Configuration

You can specify different vector database implementations through vectorDBConfig:

{
  "vectorDBConfig": {
    "type": "faiss", // Supported types include: chroma, faiss, milvus
    "host": "localhost",
    "port": 6345
  }
}

Large Language Model Configuration

Supports multiple mainstream models:

{
  "llmModel": {
    "type": "gpt-3", // Supported types include: gpt-3, llama2, alpaca
    "apiEndpoint": "https://api.example.com/v1",
    "apiKey": "your-api-key"
  }
}

MPC Server Integration

Set Up the MPC Server

Quickly set up with the @ragrabbit/mcp package:

npx @ragrabbit/mcp start --port 3001

Configure the AI Client

Configure in Claude Desktop:

Edit the claude_desktop_config.json file and add the following content:

{
  "mcpServers": {
    "my-documentation": {
      "command": "npx",
      "args": ["@ragrabbit/mcp", "start"]
    }
  }
}

Configure in ChatGPT:

Set the MPC server address through environment variables:

export MPC_SERVER=http://localhost:3001

Integration with Fumadocs

Quick Integration

Run the following command in the project root directory:

pnpm install @fumadocs/extension-rag

Configuration Example

Add the following configuration to the config.json file:

{
  "fumaIntegration": {
    "enabled": true,
    "apiKey": "your-fuma-api-key",
    "workspaceId": "your-workspace-id"
  }
}

Development Guide

Environment Setup

git clone https://github.com/ragrabbit-project/rag.git
cd rag
pnpm install

Core Component Development

Vector Database Interface

Implement a custom vector database:

interface VectorDB {
  init(): Promise<void>;
  addVectors(vectors: number[], metadata: object[]): Promise<string[]>;
  searchSimilar(id: string, k: number): Promise<SearchResult[]>;
}

Text Processing Pipeline

Define the text processing flow:

class TextProcessingPipeline {
  private tokenizer;
  private embeddings;

  constructor() {
    this.tokenizer = new Tokenizer();
    this.embeddings = new LlamaEmbedder();
  }

  async process(text: string): Promise<ProcessedText> {
    const tokens = this.tokenizer.tokenize(text);
    const embeds = await this.embeddings.embed(tokens);
    return { tokens, embeddings: embeds };
  }
}

Contribution Guide

Welcome contributions from the community! Please refer to the CONTRIBUTING.md file for the specific process.

💻 Usage Examples

Basic Usage

const rag = new RagManager({
  vectorDB: new ChromaDB(),
  llm: new Llama2()
});

async function search(query: string) {
  const results = await rag.search(query, { limit: 5 });
  return results.map(r => r.metadata);
}

📚 Documentation

Root Directory Structure

project/
├── config.json          # Global configuration file
├── package.json        # Project dependency management
├── src/                # Source code directory
│   ├── core/           # Core algorithm module
│   └── ui/             # User interface components
└── public/             # Static resource directory

Configuration File Explanation

config.json

{
  "mode": "development",      # Mode: development or production
  "vectorDBConfig": {         # Vector database configuration
    "type": "chroma",
    "host": "localhost",
    "port": 2000
  },
  "llmModel": {               # Large language model configuration
    "type": "gpt-3",
    "apiKey": "..."
  }
}

Project Maintainers

• Marco D'Alì - Project lead
• Community Contributors - List of code contributors

📄 License

This project follows the MIT open - source license. See the LICENSE file for details.

⚠️ Important Note

All examples and configurations in this document are general guidelines, and the specific implementation needs to be adjusted according to actual requirements.