rag

name	rag
description	Retrieval-Augmented Generation over your Obsidian vault. Query your knowledge base with hybrid retrieval (keyword + graph expansion via wikilinks). Inspired by LightRAG. Use when you want to ask questions about your notes, search your vault with AI synthesis, or /rag.

RAG Skill

Retrieval-Augmented Generation over your Obsidian vault. Uses Python for retrieval (ripgrep + wikilink graph) and AI only for synthesis.

Usage

python3 _scripts/rag_query.py "What do I know about AI agents?"
python3 _scripts/rag_query.py "Summarize my notes on habits" --mode hybrid --save

Options

• query — Question to answer (required)
• --mode {naive,local,hybrid} — naive: keyword only; local: keyword + wikilink expansion; hybrid: broadest (default)
• --top-k N — Max notes to retrieve (default: 12)
• --depth N — Graph expansion depth (default: 1)
• --save — Save response to vault

Output

Sources/RAG Query - YYYY-MM-DD.md (with [[wikilinks]] citations)