knowledge_search (POST /v1/knowledge/search) when you need a human-readable answer rather than raw data.
Input can be a Cala QL expression or a natural-language question — both produce the same response shape. See the Structured Answer page for the full QL syntax reference.
- Automatically identifies and extracts entities from results
- Every claim is traceable back to its source documents
Example
Answer
Answer
A succinct, structured answer in markdown — ready to use directly.
Explainability
Explainability
Every claim in the answer is traced back to specific source documents — so you know exactly why something was said.Each
reference UUID links to a source document in the Sources section below.Sources
Sources
Full provenance for every claim — publisher, article title, and URL. Nothing is a black box.
Entities
Entities
Every entity mentioned is automatically identified, typed, and assigned a Cala UUID — ready to query further with
retrieve_entity.Query language input also works
You can also send a plain question — the response shape is identical:What you get back
| Field | Description |
|---|---|
content | A succinct markdown answer with structured sections |
explainability | Reasoning chain — each claim linked to specific source documents by UUID |
context | Source documents with full provenance: publisher name, article title, URL |
entities | Every entity mentioned, with Cala UUIDs and types (Company, Organization, GPE, Facility, etc.) |