> ## Documentation Index
> Fetch the complete documentation index at: https://docs.breadbowl.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Rank without an index

> Rank a small set of documents for one request

Use `POST /v1/score` when you already have a small set of documents and only need to rank them once. The documents are not added to an index.

The example uses `BREADBOWL_API_URL` and `BREADBOWL_API_KEY` from the [quickstart](/quickstart).

```bash theme={null}
curl --fail-with-body \
  "$BREADBOWL_API_URL/v1/score" \
  -H "Authorization: Bearer $BREADBOWL_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "query": "What is the refund window?",
    "documents": [
      {"doc_id": "a", "text": "Refunds are accepted within 30 days."},
      {"doc_id": "b", "text": "Orders ship in two business days."}
    ]
  }'
```

The response lists the documents in relevance order. IDs and metadata come from your request.

Use transient scoring for a request-local candidate set of up to 50 documents. Use [indexed search](/guides/search) when documents are reused across queries or need lifecycle management and metadata filters.

BreadBowl Embed records aggregate usage for the request but does not create stored documents or indexed representations. See [data handling](/security/data-handling) for details.