[sasan1200]

Summary

• File telemetry: while the service reads enabled and launchd holds no job for the helper, the app re-registers it on each repair probe, up to 3 times a launch, even after this launch's own registration. Replacing the app in place left the old helper running at launch; the launch's re-register of it left launchd with no job, and the once-per-launch rule stopped there until a manual secure-agent telemetry repair, which loaded it at once.
• The repair probe stays one at a time: it is cleared only after its re-register, so the refresh that re-register runs cannot start a second probe.
• A refused (EX_CONFIG) or stale job is still re-registered once per launch.
• Tests: the decision table now retries a missing job after this launch's attempt and stops at the bound; a SetupManager test re-registers on each probe that finds no job, leaves a loaded job alone, and stops at 3.

Type of Change

• 🐛 Bug fix (non-breaking change which fixes an issue)
• ✨ New feature (non-breaking change which adds functionality)
• 🔒 Security enhancement
• 📝 Documentation update
• 🎨 Refactoring / Code style cleanups

Verification Checklist

• go test ./... passes cleanly
• python3 plugin/hooks/test_secret_guard.py and other python hook tests pass
• bash packaging/swift_macos.sh test --package-path menubar passes
• ./packaging/test/e2e_smoke.sh passes
• No secrets, tokens, or private keys are exposed or committed