# Managing sources, knowledge and supporting evidence

Availability: Planned API.
Last reviewed: 2026-08-28.

Keep the sources behind business decisions useful, attributable and retrievable. The planned outcome joins website evidence, documents, normalized knowledge and grounded answers without treating every raw artifact as public.

## Register the evidence scope

Provide owned source references, approved URLs or documents, the purpose, permitted use and applicable workspace/resource grants. Source verification is a prerequisite, not blanket permission for unrestricted capture.

Exact-page capture, whole-site mapping, document normalization, indexing and analysis are separate work. Each should preserve its input identity and bounded scope.

## Retain durable knowledge

Save source metadata, capture time, content/artifact references, parser or schema version, relationships and freshness. Grounded answers should point back to permitted evidence and distinguish observations from derived conclusions.

For example, a merchant specification document can support a product attribute while a conflicting page observation remains visible. A normalized file is not proof of external remediation or approval.

## Read without hidden work

Retrieving saved evidence should not crawl, refresh, embed, index or repair data. Query-time model work, if needed, must be explicit. Raw HTML, provider payloads, embeddings and signed storage addresses require separate access/retention policy.

## Availability

Business APIs are planned. User-owned knowledge records do not confer third-party copyright or redistribution rights. Preserve citations and unknowns before promoting evidence into catalogs, Stores, research or campaigns.