The Camera Health tab's "Re-check cameras" button: ask every in-scope motion_clip camera for a fresh health sample now instead of at the next hourly tick.

Exactly the tab's own scope, by construction: the same filter checks as the GET above (one shared helper), and the same TenantScope handed to the service, which bounds the write with the very predicate list the read uses. Whoever can see a camera on the tab can re-check it, and nobody can re-check a camera they cannot see.

Who may press it is NARROWER than who may read the tab: the require_camera_health_recheck role allowlist above (ops-view's canMutate) refuses the end-customer logins and read_only, which the GET admits. What an admitted caller may reach is NOT narrower: once past the role gate the scope is the tab's own, so a site manager re-checks only its granted sites. Not activity-logged: a re-check changes no configuration, it only asks for one more sample, and the service's per-camera cooldown bounds how many.

{
  "parent_id": "11111111-1111-1111-1111-111111111111",
  "site_id": "33333333-3333-3333-3333-333333333333"
}

{
  "requested": 12,
  "already_pending": 3,
  "unsupported": 2,
  "cooldown_seconds": 120
}