bigRAG
API Reference

Evaluation

Measure retrieval quality against a labelled query set.

POST /v1/evaluation

Runs a batch of queries against a collection and scores the retrieved chunks against a labelled set of expected document IDs. Used to catch regressions when tweaking chunk size, embedding model, reranking, or search mode.

Accepts a session cookie or an API key with query:read.

The admin UI exposes the same runner under Settings → Evaluation.

Usage

Aggregate per-collection usage statistics.

User Preferences

Per-user JSON blob for admin UI state.