You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
test: close macOS platform command-coverage gaps from #1426 #1916
The macOS platform manifest records these commands as known gaps because HEAD has no command-specific macOS evidence that meets the manifest contract:
doctor
shutdown
events
replay
test
batch
diff
react-native
trace
find
fill
longpress
type
focus
swipe
Follow-up work should add or identify executable macOS command evidence before changing the manifest rows. Keep the existing macOS replay and provider scenario scope unchanged unless a separate platform-evidence decision explicitly expands it.
Decision: grow the macOS live scenarios to cover the interaction gaps: find, fill, type, focus, longpress, swipe, diff and batch, through the System Settings replay or the desktop provider scenario. doctor, shutdown, events, test, trace and react-native stay declared gaps.
Correction to the decision above: test stays owed work here, not a declared gap. The macOS replay gate already runs agent-device test … --report-junit and fails on suite failure, so that row can close with existing evidence. #3717 records doctor, shutdown, events, trace and react-native as accepted limitations with reasons. find, fill, type, focus, longpress, swipe, diff, batch, replay and test stay on this issue.
Activity
thymikee commentedon Oct 16, 2026
Decision: grow the macOS live scenarios to cover the interaction gaps: find, fill, type, focus, longpress, swipe, diff and batch, through the System Settings replay or the desktop provider scenario. doctor, shutdown, events, test, trace and react-native stay declared gaps.
test(coverage): keep macOS test-suite evidence as owed work and audit…
thymikee commentedon Oct 16, 2026
Correction to the decision above:
teststays owed work here, not a declared gap. The macOS replay gate already runsagent-device test … --report-junitand fails on suite failure, so that row can close with existing evidence. #3717 records doctor, shutdown, events, trace and react-native as accepted limitations with reasons. find, fill, type, focus, longpress, swipe, diff, batch, replay and test stay on this issue.test(coverage): split gap rows into owed work and accepted limitations (