Knowledge sources for the investigator

Embed
Preview of the card
<a href="https://watchtowerops.net/docs/ai-knowledge"><img src="https://watchtowerops.net/docs/_api/public/orgs/aa5959cb-1a0b-4a9a-bc9a-913e7ac31cd9/spaces/b7aac181-9916-4805-b253-becc5fd2ca62/pages/f9449840-0589-44d6-8524-ce04eb761c4d/social-card.svg?v=3e7ae33eec8b" alt="Knowledge sources for the investigator" width="600" height="315"></a>

Documents the investigator can search beside runbooks. An exporter, a CI job or a person pushes text with PUT /v1/projects/{prj}/knowledge/docs (needs ManageAi); it is chunked and full-text indexed.

Field

Meaning

source

Exactly one of confluence, notion, gdocs, repo, upload, other. Anything else is a 400.

external_id

The page id or file path at the source; with source it identifies the document, so a second PUT replaces it.

title

Document title.

url

Optional https link to the original.

text

Plain text or markdown, at most 512 KiB.

GET .../knowledge/docs lists documents, DELETE .../knowledge/docs/{doc} removes one, and GET .../knowledge/search?q=&limit=&cursor= searches (q is web-style: words, "a phrase", -exclude, or; 2 to 200 characters; limit 1 to 20, default 5). Hits carry doc_id, title, url, source, snippet and rank. Documents are data: the investigator treats their text as untrusted.

Updated