Every verdict comes from a measured signal and a written rule: 56 rules. The display rules are replayed against the real venue log in the test suite, and 28 of the 56 are named in a test by id. 38 of the rules name a failure and carry their own evidence and fix; the rest say why a row is GO, or what it has to ask you. A check that cannot measure its signal says SKIPPED, never GO. A screen that has been up for less than two minutes is CHECKING, never GO.
The beta has no AI, no account and no analytics. AI help comes later, and only explains: it will read the same report you can see here, put it in plain words, and never change a verdict.
This is all a report holds: statuses, numbers and rule ids. It is a real one, from the Mac this site was built on, at 2026-11-15 23:44.
- No screen names, beyond a 3-letter maker code
- No Bluetooth or audio device names
- No Wi-Fi names or IP addresses
- No URLs or hostnames
- No file names or app names
- No screenshots, audio or log lines
{
"schema": 1,
"app": "0.1.0",
"os": "27.2",
"chip": "Apple M2 Pro",
"profile": "presenter",
"verdict": "warn",
"checks": [
{
"id": "display.external",
"modes": [
"DV01"
],
"rule": "display.none",
"signals": {
"display.churn.removals300s": 0,
"display.external.count": 0,
"kernel.hdcp.aborts300s": 0,
"kernel.hdcp.aborts60s": 0,
"kernel.hpd.highHeldSeconds": 0,
"kernel.hpd.portsHigh": 0,
"kernel.irqhpd.count60s": 0,
"kernel.read.ok": true
},
"status": "ask"
},
{
"id": "audio.output",
"modes": [],
"rule": "audio.go",
"signals": {
"audio.output.muted": false,
"audio.output.present": true,
"audio.output.transport": "builtIn",
"audio.output.volumePercent": 88
},
"status": "go"
},
{
"id": "power.sleep",
"modes": [
"PW04"
],
"rule": "pw04.soon",
"signals": {
"screensaver.idleMinutes": 20,
"sleep.assertionHeld": true,
"sleep.displayMinutes": 5,
"sleep.heldByShowMode": false
},
"status": "warn"
}
]
}