ReplayMac ReplayMac QA · for software teams
The bug already happened. So has the recording.
ReplayMac QA keeps a rolling buffer of the screen while your testers work. When a tester presses a hotkey, a test runner reports a failure, or a watched app crashes, it saves the footage that led up to it — as a structured evidence package, ready to attach to the ticket.
Three triggers, zero guesswork
Every capture comes from an explicit, deterministic trigger. There is no AI watching the screen deciding what "looks wrong".
A tester presses the hotkey
Save Bug Evidence is a menu item, a hotkey, and a Settings button. The tester sees the bug, presses the key, and keeps testing — the preceding 30 seconds to 5 minutes are packaged in the background.
A test runner reports a failure
Scripts and CI-adjacent test frameworks trigger the same workflow through replaymac://qa/save or the replaymac-qa CLI — labelled, rate-limited, and never able to overlap or corrupt a capture.
A watched app crashes
ReplayMac QA watches macOS diagnostic reports for the apps you select. When a crash report appears, it saves the footage before the crash — default 2 minutes — and copies the .ips report into the package.
The evidence package
One folder per capture, assembled atomically — a package you can see is a package that's complete. Attach it to the ticket and close the loop.
Checkout-Test-Failure-2026-07-12-14-35-22/ ├── replay.mp4 the footage before the trigger ├── metadata.json machine-readable environment details (schema v1) ├── summary.txt human-readable — paste it straight into the ticket └── crash-report.ips for crash captures, if enabled
↑ lands in ~/Movies/ReplayMac/QA/ · labels are sanitised · collisions never overwrite existing evidence
Spec sheet
The details, without the marketing voice.
- Triggers
- Hotkey / menu item, replaymac://qa/save URL scheme, replaymac-qa CLI, and crash-report monitoring for watched apps.
- Capture window
- 30 seconds to 5 minutes before the trigger (default 90 s; crash captures default to 2 minutes).
- Reliability
- Captures run through a single-save gate — exports never overlap. Crash triggers outrank queued automation, automated triggers are rate-limited, and manual captures are never silently dropped.
- Crash capture
- Watched apps are identified by bundle id, with a one-time sandbox grant for the diagnostic-reports folder. The same report is never packaged twice, even across relaunches.
- Isolation
- Installs as ReplayMac QA.app with its own bundle id — settings, permissions, and buffers fully separate from a consumer ReplayMac on the same machine.
- Privacy
- No cloud account, no telemetry, no uploads — evidence stays on the Mac until you share it. Safer defaults on first run: system audio and microphone off.
- Pilot
- 14 days, starting when you enable QA mode — not when you download. Trial state is stored locally; nothing is sent anywhere.
- Requirements
- macOS 15 or later, and Screen Recording permission — that's the whole point.
What it deliberately doesn't do
- No AI bug detection. It doesn't watch the recording and guess whether something "looks wrong". Every capture is explicit and reproducible.
- No uploads, anywhere, ever. Evidence leaves the Mac only when a human decides to share it.
- No DRM theatrics. The pilot stores its dates locally and trusts you. When it ends, new captures stop — but nothing you captured is locked, deleted, or held hostage.
Start a pilot
Get a 14-day pilot build for your team, try it against real bugs, and tell me whether the evidence packages actually closed tickets faster. Pricing and licensing will be decided from pilot feedback — that's the honest state of it.
Honest small print: the pilot clock starts when you enable QA mode, not when you download. When it runs out, new captures stop — but every package you made stays yours, and the consumer replay features keep working. Existing evidence is never deleted or locked.