Once investigations are enabled, Nitpick files issues and sends notifications autonomously. In the production deployment that produced 1,030 issues, that is the desired behaviour -- but a team evaluating Nitpick on a live service has no way to review output before it reaches their tracker.
Scope
- An optional hold state where investigations complete and are visible in the dashboard, but issue creation, notifications, and fix PRs wait for approval.
- Approve and discard actions in the dashboard, per investigation.
- Configurable globally and per service, reusing the existing settings resolution (
resolve_service_settings) so it layers the same way fixes_enabled does.
- Default off, so existing deployments are unaffected.
Why it matters
This lowers the cost of trying Nitpick on a real service, which is currently the biggest adoption barrier: the first run against production either does nothing useful or writes to a live tracker.
Done when
With the gate enabled, a full run produces investigations and no external side effects until approved, and approving one performs exactly the actions it would have taken unattended.
Once investigations are enabled, Nitpick files issues and sends notifications autonomously. In the production deployment that produced 1,030 issues, that is the desired behaviour -- but a team evaluating Nitpick on a live service has no way to review output before it reaches their tracker.
Scope
resolve_service_settings) so it layers the same wayfixes_enableddoes.Why it matters
This lowers the cost of trying Nitpick on a real service, which is currently the biggest adoption barrier: the first run against production either does nothing useful or writes to a live tracker.
Done when
With the gate enabled, a full run produces investigations and no external side effects until approved, and approving one performs exactly the actions it would have taken unattended.