coyote/docs/testing/plans/12-rag.md

Test Plan: RAG

Behaviors to test

• Rag::init creates new RAG with embedding model (requires LLM client)
• Rag::load loads existing RAG from disk (requires filesystem)
• Rag::create builds vector store from documents (requires embedding model)
• Rag::refresh_document_paths updates document list (requires filesystem)
• RAG search returns relevant embeddings (requires embedding model)
• RAG template contains required placeholders
• Reranker model applied when configured (requires LLM client)
• top_k controls number of results (requires embedding model)
• RAG sources tracked for .sources command (requires full Rag struct)
• exit_rag clears RAG from context (tested in iteration 8)

Additional behaviors tested

• DocumentId: new/split round-trip, zero/zero, large values
• DocumentId: Debug format ("file-doc"), equality, inequality, ordering
• RagDocument: new with content, default empty
• RagData: new sets all defaults, empty collections
• RagData::get: returns document, None for missing file, None for missing doc index
• RagData::del: removes files + associated vectors, noop for nonexistent
• RagData::add: inserts files, vectors, updates next_file_id
• RagData::build_bm25: empty data returns no results
• RagData::build_bm25: finds documents by keyword (BM25 ranking)
• RAG_TEMPLATE: contains CONTEXT, SOURCES, INPUT
• get_separators: Rust/Python/Markdown return language-specific
• get_separators: unknown extension returns defaults
• get_separators: all 22 known extensions have language-specific separators

Old code reference

• src/rag/mod.rs — Rag struct and methods
• src/config/request_context.rs — use_rag, edit_rag_docs, rebuild_rag