WebRTC / standalone diagnostic
Why does the offer probe fail?
The sampled Chrome failed all three runs with probe_error. Stealthium retained all three. That label does not tell us whether WebRTC failed or Introspection discarded its evidence.
Runs automatically using the six production cases. No camera, microphone, remote peer or saved session. Offers are never installed as local descriptions. Run again is click-started; the initial run is page-started.
Waiting for JavaScript. No probes have run.
Leading hypothesis: evidence overflow
The collector repeats codec details, repair mappings and SDP codec lines across cases. Above 32 KiB, it packages the report as base64, increasing its size by roughly one third. The production wrapper allows only 43,008 bytes. A larger valid codec report can therefore become a generic probe error.
The sampled Chrome advertises H.264 high-profile and H.265 receive profiles absent from Stealthium. This could increase report size, but Chrome’s failed capture retained no offers or original exception. Three failures establish repeatability in that capture, not that genuine Chrome always fails.
Waiting for native outcomes and retention checks.
Failure boundary
The invented codec should be rejected at the setter. Successful native cases followed by a wrapper size failure implicate evidence handling. A native failure needs its own investigation.
| Check | Current browser |
|---|
Collect evidence for comparison
Download one payload from each browser, whether it overflows or passes. It includes the original case evidence, exact production wrapper, byte breakdown, build identity, browser and OS claims, video capabilities and graphics renderer. Extra system queries run after the offer probe finishes.
These notes are supplied context. Browser scripts cannot verify these settings. Export each run before rerunning; the page retains only the latest completed capture.
No export yet.
Inspect or manually copy JSON
Automated collection: ?json=true returns the same evidence schema. Browser and GPU strings are reported claims, not proof of the native codec backend.
A useful reasoning signal is capability → preference acceptance → offered codecs consistency, with retention failures reported separately. This page diagnoses the failure boundary; it does not classify a browser as genuine or modified. Open the full signal.