tya5/rag

v0.1.0

Builtin turnkey user RAG (ADR 0064 P5): chunker + vector-store MCP servers, ingest/query pipelines, and its RAG skill (install, embedding-provider setup, the ingest/query workflow, and corpus internals, all bundled with references). External sqlite store the operator names and owns (FP-0057 C2) -- distinct from reyn's own OS-internal index (not agent-callable, FP-0066 P1b).

GitHub

build-and-query-rag-corpus

Make a folder of the operator's own documents (txt/md/pdf/xlsx/pptx/docx) searchable by meaning via a user-named sqlite vector store. Covers installing the rag plugin, embedding-provider setup, the exact rag_ingest/rag_query pipeline calls, and corpus internals -- via bundled references. Read this before running the builtin rag_ingest / rag_query pipelines, or when the operator asks you to search a folder/file of their own documents.