Device Lab run guides

You can repeat a Device Lab test on an iPhone or iPad using the protocol published with the 2026-10-20 study, then compare your result with the published data. The only published run is on one iPhone; the protocol plans an iPad subset if time allows.

Guides

• Run the tests on iPhone and iPad: the whole run from the hash check to closing the run.
• Set up Web Inspector: pair an iPhone with a Mac and read page state from Safari tabs and Home Screen apps.
• Screenshots and recordings with file names: the naming rule and environment block, with instructions for redaction.
• Send your results: copy the JSON a harness page shows and email it.

Rules for every run

Rules 1 and 2 and rules 3 and 5 come from the published protocol. Rule 4 follows the reproducibility notes in the study report (section 13).

1. Record a push subscription’s endpoint host only (for example web.push.apple.com), never the full endpoint or its keys. Keep other apps’ content out of the evidence: crop or skip any screenshot that shows another app’s notification, a contact, the carrier name or the location. Remove identifiers such as the device UDID before you publish or send a file; the study data package replaced the UDID in its read script with a variable before publication.
2. Fill in an environment block once per device per session: date and time with the time zone; device marketing name and model number; OS version and build; release channel; Safari version and user agent; device language and region; any Safari feature flag changed from its default; content blockers and VPN, plus iCloud Private Relay; the Mac and Safari version used for Web Inspector, if any.
3. When a test asks for it, remove the Home Screen icon (touch and hold, then choose Remove App followed by Delete) and clear the site’s data in Settings > Apps > Safari > Advanced > Website Data before you start.
4. Compare the files you load with the hashes in /lab/harness/v1/SHA256SUMS.txt, which lists the SHA-256 of each harness v1 file as served. The report asks for this because the lab can change: during the 2026-10-20 run a deploy at about 08:28 UTC changed lab.js, and pages that were already open kept the cached earlier copy.
5. Leave Safari’s feature flags as they are. Record any that are already non-default.

Other platforms

The 2026-10-20 study ran on one iPhone. Android Chrome and desktop Chromium have no run guide in this phase; neither do Firefox and Samsung Internet. The roadmap lists those tests: share target and install prompts on Android, file handling and launch handling on desktop Chromium, Firefox install surfaces, and storage persistence.

Device steps performed by an OpenPWA operator on an iPhone 16 Pro. AI agent sessions read page state over USB, kept the notebook and drafted this text; automated gates and an AI fact-check checked it. No person reviews each entry.