Ingest documents into a named collection so they can be searched.

Provide paths (files or directories of .txt/.md) and/or texts (raw strings). Documents are chunked, embedded, and indexed for both semantic and keyword search. Returns how many chunks were added and the collection's new total.

Search a collection and return the most relevant chunks, each with its source.

mode is "hybrid" (BM25 + semantic, default), "dense" (semantic only), or "bm25" (keyword only). rerank applies the cross-encoder reranker if one is configured. Each result includes per-stage scores for transparency. An empty list means nothing relevant was found — treat that as "not in the documents".

Answer a question grounded in a collection, with citations.

Retrieves the most relevant passages and asks the host's model (via MCP sampling) to answer using ONLY those passages, citing them. Returns {grounded, answer, citations}. If nothing relevant is found, grounded is false and no answer is invented. If the host does not support sampling, the grounded context is returned for the host to compose the answer itself.

List all ingested collections and how many chunks each contains.

Measure retrieval quality on labeled cases: hit_rate, MRR, and recall@k.

Each case is {query, relevant_sources}. Use this to quantify quality and catch regressions — e.g. before and after changing chunking or switching embedders.