Skip to main content

VRChat claiming verification

Keep three kinds of evidence separate:

  • Collector availability: read communityTelemetry:collectorProofAvailable and exact-release collectorDeploymentReadiness. These establish that the worker is eligible and polling, not that a claimant's code was found.
  • Isolated authorization coverage: the backend test below uses two identity fixtures and synthetic collector verdicts against real application handlers. The browser test uses two real Clerk development accounts and the existing staging adapter fixture. Neither establishes live VRChat identity proof.
  • Live ownership proof: an authorized claimant places the generated code in the target bio or group description and the collector reads it successfully. Do not substitute a fixture verdict or edit another person's account to manufacture this evidence.

Automated two-user checks​

Run the isolated backend lifecycle:

node --conditions=import --import tsx --test tests/backend/vrchat-two-user-claims.test.ts

It checks distinct codes, private attempt context, reciprocal rejection of another user's attempt ID, cancellation isolation, a competing ownership result, and replay safety. It asserts a single owner, control proof, external link, and approved claim request, with no loser-owned side effects. Network calls fail the test.

For the browser check, use the existing hosted Clerk development setup. Enable the already configured staging auth and adapter helper flags in the test process, then run:

pnpm --filter web exec playwright test e2e/auth-claim.flow.spec.ts --grep "two Clerk users" --project=desktop-chromium

This creates two disposable Clerk accounts, opens separate browser contexts, and creates one run-attributed profile. Both users obtain private codes. User B cancels their attempt; user A's survives reload and completes through the staging adapter fixture. User B then sees the existing-owner state, while user A retains management access after reload. Both accounts and the fixture profile are cleaned up. A configured run fails if its adapter prerequisite is missing; local auth-unavailable runs skip and must not be reported as coverage.

Do not enable helpers in production. Use separate run IDs and browser contexts from other staging work. No additional auth bypass or shared helper changes are needed for this test.

Remaining live check​

Use an existing authorized claimant and a real pending attempt. Confirm the collector is available, let the claimant publish the code, and verify the attempt's successful completion and resulting ownership. A not_found result demonstrates a working provider read, but leaves ownership unproven. The collector runbook documents NAT-origin session recovery if the provider rejects the collector session.