Expand description
Turbopuffer vector-store backend for Knowledge Indexes.
Implements everruns_contracts::vector_store::VectorStore against Turbopuffer’s
v2 HTTP API (https://turbopuffer.com/docs). This is the reference
production backend; the in-memory store stays the default and Turbopuffer is
opt-in via TURBOPUFFER_API_KEY.
Part of the Everruns ecosystem.
use everruns_turbopuffer::TurbopufferVectorStore;
let _store = TurbopufferVectorStore::new("https://example.turbopuffer.com", "api-key");Multitenancy and multi-index isolation are handled by the caller: each
Knowledge Index gets one org-prefixed namespace (see
everruns_contracts::vector_store::index_namespace), and this backend simply
maps each namespace string onto a Turbopuffer namespace path segment.
§Error handling
Non-2xx responses are mapped to sanitized anyhow errors. The API key is
only ever sent in the Authorization header and is never included in any
error message or log line.
Structs§
- Turbopuffer
Vector Store - Turbopuffer-backed
VectorStore. - Usage
- Provider-reported resources, before monetary pricing. Missing receipts must never be replaced with guessed request sizes or a fabricated query count.