This project provides a local, UI-driven ecosystem for GraphRAG, enabling users to build and query knowledge graphs using local LLMs. It targets developers and researchers seeking a cost-effective, self-hosted solution for advanced RAG applications, offering a comprehensive suite of tools for indexing, prompt tuning, querying, and visualization.

How It Works

The system is built around a FastAPI-based core API that orchestrates all GraphRAG operations. It supports local LLMs and embedding models via Ollama or OpenAI-compatible APIs, eliminating cloud dependencies. A separate Gradio UI handles indexing and prompt tuning, while a legacy UI offers visualization and querying, with plans for a dedicated querying/chat UI. This modular, API-centric architecture promotes flexibility and maintainability.

Quick Start & Requirements

• Install: pip install -e ./graphrag followed by pip install -r requirements.txt.
• Prerequisites: Python 3.x, Conda environment recommended. Ollama or OpenAI-compatible API for local models.
• Run: Start API (python api.py), optionally embedding proxy (python embedding_proxy.py), then UIs (gradio index_app.py, gradio app.py).
• Docs: Original GraphRAG paper, GraphRAG documentation.

Highlighted Details

• API-centric architecture with FastAPI.
• Dedicated Gradio UIs for indexing/prompt tuning and a legacy UI for querying/visualization.
• Local LLM and embedding model support (Ollama, OpenAI-compatible).
• Real-time 2D/3D knowledge graph visualization via Plotly.
• File management, settings management, and real-time logging within the UI.

Maintenance & Community

The project is actively developed, though updates have been slowed by the author's day job. Contributions and PRs are encouraged. Users can report issues and request features via GitHub Issues.

Licensing & Compatibility

The project's license is not explicitly stated in the README. Compatibility for commercial use or closed-source linking is not specified.

Limitations & Caveats

The project is undergoing a major transition with separate UIs being developed, potentially leading to instability. It has been primarily tested on macOS (M2 Studio), and Windows users may encounter encoding issues. Support for additional file formats and advanced graph analysis tools is planned but not yet implemented.