ai-friendly-docs

This project transforms official documentation from open-source projects like Angular into formats optimized for consumption by Large Language Models (LLMs). It enhances AI response quality and accessibility by creating structured markdown files and generating vector embeddings for semantic search. The processed documentation serves as high-quality training data and improves Retrieval-Augmented Generation (RAG) systems with accurate technical content.

Key Features

• Repository Cloning — Automatically clones specific versions of source repositories from GitHub.
• Documentation Generation — Transforms original documentation into AI-consumable markdown formats, including comprehensive single files and individual feature sections.
• Vector Embedding Creation — Generates embeddings using HuggingFace models and stores them in a Qdrant vector database for semantic search.
• Structured Output — Maintains organized directory structures with complete documentation, table of contents, and version-specific sections.
• RAG Integration — Provides ready-to-use vector stores for enhancing LLM responses with accurate documentation retrieval.

Philosophy

The project believes that documentation structured for AI consumption significantly improves LLM understanding and response accuracy, bridging the gap between human-readable content and machine learning efficiency.