Change the evaluation settings

PATCH 
/api/knowledge/evaluation/settings

auto_eval (doc 103 §12, default on): when the weekly knowledge health check finds no gold set, it queues one evaluate operation that writes the test questions and scores them — only with the AI switched on, usable embedding and chat models and, on the platform key, a wallet that covers the estimate; at most once a week. Every decision is written to the health check's operation log. Requires knowledge management.

Request

Responses

200

The settings now in force.

401

Not authenticated. code: account_disabled means the member's seat was deactivated in this workspace: the session is over even though the token was not revoked (clients sign out and clean up).

403

Authenticated but not allowed.

422

The request failed validation.