Knowledge

The knowledge base is what the agent knows beyond the conversation. Ingest URLs and sources; they're processed and made retrievable so missions can ground their content in real material.

Knowledge is a general-purpose base layer — the unstructured half (the structured half is records). A source can be scoped to:

• a subject (subjectId) — material about a person/brand/product,
• a mission (missionId) — material for one promotion,
• a project (projectId) — e.g. reference docs for a research dataset,
• or nothing — a workspace-level source the whole agent can draw on.

Pass whichever scope fits (or none) to knowledge_ingest_url; missions and research initiatives both sit on this same layer.

Use in the UI

Open Knowledge under Settings (/knowledge).

• Ingest a URL or source; refresh it to re-pull.
• Browse and edit ingested entries.

Use with AI

• knowledge_ingest_url, knowledge_refresh
• knowledge_list, knowledge_get, knowledge_update, knowledge_delete
• knowledge_search — semantic retrieval across ingested material.

The agent ingests reference material with knowledge_ingest_url, then pulls from it via knowledge_search while drafting.

Use the API

curl -X POST https://<host>/api/invoke/knowledge.ingest_url \
  -H "Authorization: Bearer $NIHMBUS_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "url": "https://example.com/brand-guidelines" }'

Reference

Knowledge — includes knowledge.search for semantic retrieval.