Introduce a lightweight local Retrieval-Augmented-Generation (RAG) toolchain for the Aetheria Obsidian vault. Added scripts/rag/rag.py (indexing, chunking, embedding, hybrid semantic+lexical query, stats, and install), a PowerShell wrapper scripts/rag/rag.ps1 that resolves Python and forwards commands, and scripts/rag/README.md with usage (install, build, query, stats). Add requirements-rag.txt for qdrant-client[fastembed] and pywin32, and include a compiled __pycache__ file. Update .gitignore to ignore the .rag/ artifact directory and update AGENTS.md with Vault Navigation guidance recommending the local RAG workflow. The tool stores local artifacts under .rag/ (pydeps, qdrant, model cache, chunks and manifest) and defaults to a CPU-friendly embedding model.
f339aae186