Release v0.1 (12/13): Clean-machine verification (pre-release QA)

[fdiotalevi]

Step 12 of the v0.1 release sequence. — pre-release QA.

Verify the real user experience on a machine that never had the repo, before publishing.

Tasks

• On a clean Mac (or fresh user account / VM): install from the notarized DMG produced by make_release.sh.
• Confirm Gatekeeper accepts it with no warning (a machine where it was never built/signed locally).
• Complete the 3-permission onboarding (Microphone, Input Monitoring, Accessibility) from scratch.
• Smoke-test all features: dictate, Recent Dictations, re-paste last dictation, and the Configuration panel (record a shortcut, save, confirm it persists).

Acceptance

• The DMG installs and the app runs end-to-end on a clean machine with no developer setup.

Depends on steps 9, 10, 11.

[fdiotalevi]

Clean-machine pre-release QA completed and passed.

QA surfaced 3 macOS TCC onboarding bugs in the frozen .app, all fixed and shipped in v0.1.0 (a122f83):

• Microphone prompt was silently auto-denied — added the hardened-runtime com.apple.security.device.audio-input entitlement.
• A post-launch Accessibility / Input Monitoring grant was never detected (frozen sys.executable rejects -c) — added an internal --preflight flag.
• System Settings deep-links didn't open from the background app — switched from the regressed open CLI to NSWorkspace.openURL_.

Verified end-to-end: notarized DMG installs, Gatekeeper accepts with no warning, 3-permission onboarding completes from scratch, and dictation works (confirmed by a live dictation through the full record → transcribe → paste pipeline).