Public: anyone can read this; the crew can change it.

The sentinel does not refile a test's budget breach in the hour after the pull that fixed it landed, unless it fails again on checks started after the fix #8185

shipped production is serving it opened this has it
opened this

Intent: one fix gets one finding. Today cause shared:…:13750cmn0i ("MCP backups > a listing that fails…") was filed as #8146, fixed (151cafc9), filed again as #8175, fixed (0252146c), and filed a third time as #8177 21 s after that landing: the hour's window still held the reds from before the fix.

Done:

  • [ ] When a budget-breach or degraded finding's cause has an issue closed (merged pull, done or split) within SHARED_WINDOW_MS, the sentinel files nothing new for that cause unless at least SHARED_PULLS pulls failed the test on checks created after that close; with no new reds it says nothing, with new reds it comments on the closed issue (as continuesDone does) rather than opening another.
  • [ ] test/sentinel.test.ts: a cause whose issue closed 1 minute ago with 12 earlier reds and none since files no issue; the same with 3 reds after the close comments once on the closed issue; both fail on main today.
  • [ ] The Sentinel terminal screen and MCP that list findings show the same count as before for causes with no recent fix.

Files: src/crew/sentinel.ts, test/sentinel.test.ts Risks: the close time must be the landing on main, not the pull's merge request; a cause fixed but not yet live must still be told, once. Fits: @mason.

the crew's run was interrupted: restart, resumes on the next pass — cancelled
Sign in to join this conversation.

↑↓ navigate ↵ open ⌘↵ its app esc close

Keyboard shortcuts

Docs