Expand description
§spacedb-vector — SpaceDB Layer 4 (the private-RAG substrate)
A native, on-node vector index co-located with the data: a query embedding
goes in, and only the top-k results come out — the corpus never leaves the
home. An authorized AI agent’s retrieval runs here, gated by an mID
capability (retrieve), which is the concrete mechanism behind
“inaccessible by default, accessible by mID-gated consent”: semantic access to
results, not bulk data.
Open-core (MIT): depends only on spacedb-access (for the capability gate).
S3 ships an exact flat k-NN VectorIndex; sub-linear ANN (HNSW / IVF) is a
future optimization behind the same surface.
Structs§
- Match
- A search hit: which entry, and how similar. No vector is returned — the corpus stays on the node.
- Vector
Index - An on-node embedding index.
Enums§
- Metric
- How similarity is measured.
searchalways returns the highest-scoring entries, so every metric is oriented “higher = more similar”. - Vector
Error
Functions§
- retrieve
- Retrieve the top-
kmatches forqueryfromindex, but only ifauthorizationallows (e.g. an agent presenting acompute-scoped capability). Denied callers getVectorError::Deniedand no data.