Vector Search

Upsert vectors

curl -X POST https://magnusdb.dev/api/vector/upsert \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "id": "doc-001",
    "text": "Magnus is a multi-model database for Cloudflare Workers",
    "metadata": {"type": "documentation", "category": "overview"}
  }'

Magnus auto-generates embeddings using bge-base-en-v1.5 via Workers AI — you send text, not vectors.

Query

curl -X POST https://magnusdb.dev/api/vector/query \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "query": "how does routing work",
    "topK": 10,
    "filter": {"type": "documentation"}
  }'

Returns the top-K most semantically similar results with scores and metadata.