name: vector-store
description: Transient vector store service for fast similarity search (FAISS)
allowed-tools: curl
commands:
- name: serve
description: Start the vector store service
usage: ./run.sh serve
- name: sanity
description: Run sanity tests
usage: ./sanity.sh
metadata:
port: 8600
endpoints:
- /index
- /search
- /reset
- /info

Vector Store Skill

Transient vector store service for fast similarity search using FAISS (IndexFlatIP).
Designed to be used by other skills (memory, edge-verifier) to accelerate KNN search.

Configuration

• VECTOR_STORE_PORT: Port to listen on (default: 8600)

API

POST /index

Add vectors to the index.

{
  "ids": ["id1", "id2"],
  "vectors": [[0.1, 0.2, ...], [0.3, 0.4, ...]],
  "reset": false
}

POST /search

Search for nearest neighbors.

{
  "query": [0.1, 0.2, ...],
  "k": 10
}

Returns:

{
  "ids": ["id2", "id1"],
  "scores": [0.95, 0.88]
}

DELETE /reset

Clear the index.

GET /info

Get index stats.