Fast Godot RAG
Queries and interprets Godot engine official documentation using a retrieval-augmented generation approach.
About
This command-line assistant leverages Retrieval-Augmented Generation (RAG) to provide intelligent answers by querying and interpreting Godot engine's official documentation. It features a robust pipeline that includes converting raw `.rst` documents into structured Markdown, intelligently splitting these into RAG-suitable JSON fragments, and then vectorizing them using Hugging Face semantic models. A FAISS vector database is built locally for efficient information retrieval, and the system integrates with the DeepSeek Chat API via FastMCP tools, enabling structured and interactive Q&A directly from a personalized Godot documentation knowledge base.
Key Features
- Automated `.rst` to structured Markdown conversion
- Intelligent Markdown splitting for RAG-optimized chunking
- Semantic vectorization using Hugging Face models
- Local FAISS vector database for efficient indexing and retrieval
- FastMCP tool registration for API-driven interaction
- 3 GitHub stars
Use Cases
- Efficiently querying specific information within Godot engine documentation
- Gaining deeper understanding and interpretation of Godot engine concepts
- Building an interactive Q&A interface for local Godot documentation
