LightRAG/README.md

🚀 LightRAG: Specialized Production Fork

A production-ready fork of LightRAG featuring S3 storage integration, a modernized Web UI, and a robust API.

🔱 About This Fork

This repository is a specialized fork of LightRAG, designed to bridge the gap between research and production. While preserving the core "Simple and Fast" philosophy, we have added critical infrastructure components:

• ☁️ S3 Storage Integration: Native support for S3-compatible object storage (AWS, MinIO, Cloudflare R2) for scalable document and artifact management.
• 🖥️ Modern Web UI: A completely redesigned interface featuring:
  • S3 Browser: Integrated file management system.
  • File Viewers: Built-in PDF and text viewers.
  • Enhanced Layout: Resizable panes and improved UX.
• 🔌 Robust API: Expanded REST endpoints supporting multipart uploads, bulk operations, and advanced search parameters.
• 🛡️ Code Quality: Comprehensive type hinting (Pyright strict), Ruff formatting, and extensive test coverage for critical paths.

---

📖 Introduction to LightRAG

LightRAG incorporates graph structures into text indexing and retrieval processes. This innovative framework employs a dual-level retrieval system that enhances comprehensive information retrieval from low-level entities to high-level broader topics.

Algorithm Flowchart

Figure 1: LightRAG Indexing Flowchart (Source)

⚡ Quick Start

1. Installation

This project uses uv for fast and reliable package management.

Option A: Install from PyPI

uv pip install "lightrag-hku[api]"

Option B: Install from Source (Recommended for this Fork)

git clone https://github.com/YourUsername/LightRAG.git
cd LightRAG
uv sync --extra api
source .venv/bin/activate

2. Running the Server (UI + API)

The easiest way to experience the enhancements in this fork is via the LightRAG Server.

1. Configure Environment:

   cp env.example .env
   # Edit .env to add your API keys (OpenAI/Azure/etc.) and S3 credentials
   

2. Start the Server:

   lightrag-server
   

3. Access the UI: Open http://localhost:9600 to view the Knowledge Graph, upload files via the S3 browser, and perform queries.

3. Python API Example

You can also use LightRAG directly in your Python code:

import os
import asyncio
from lightrag import LightRAG, QueryParam
from lightrag.llm.openai import gpt_4o_mini_complete, openai_embed

WORKING_DIR = "./rag_storage"
if not os.path.exists(WORKING_DIR):
    os.mkdir(WORKING_DIR)

async def main():
    # Initialize LightRAG
    rag = LightRAG(
        working_dir=WORKING_DIR,
        embedding_func=openai_embed,
        llm_model_func=gpt_4o_mini_complete,
    )
    await rag.initialize_storages()

    # Insert Document
    await rag.ainsert("LightRAG is a retrieval-augmented generation framework.")

    # Query
    print(await rag.aquery(
        "What is LightRAG?",
        param=QueryParam(mode="hybrid")
    ))

if __name__ == "__main__":
    asyncio.run(main())

📦 Features & Architecture

Storage Backends

LightRAG now targets a single, production-grade stack: PostgreSQL with pgvector and AGE-compatible graph support. Object storage remains pluggable (S3 or local).

Type	Implementations
KV Storage	PGKVStorage
Vector Storage	PGVectorStorage (pgvector)
Graph Storage	PGGraphStorage (AGE/PG)
Object Storage	S3Storage (New), LocalFileStorage

Specialized API Routes

This fork exposes additional endpoints:

• POST /documents/upload: Multipart file upload (supports PDF, TXT, MD).
• GET /storage/list: List files in S3/Local storage.
• GET /storage/content: Retrieve file content.

🛠️ Configuration

See env.example for a complete list of configuration options. Key variables for this fork:

# S3 Configuration (Optional)
S3_ENDPOINT_URL=https://<accountid>.r2.cloudflarestorage.com
S3_ACCESS_KEY_ID=<your_access_key>
S3_SECRET_ACCESS_KEY=<your_secret_key>
S3_BUCKET_NAME=lightrag-docs

📚 Documentation

• API Documentation
• Offline Deployment
• Docker Deployment

🤝 Contribution

Contributions are welcome! Please ensure you:

1. Install development dependencies: uv sync --extra test
2. Run tests before submitting: pytest tests/
3. Format code: ruff format .

📜 License

This project is licensed under the MIT License. See the LICENSE file for details.

---

_{Fork maintained by clssck. Based on the excellent work by the HKUDS team.}