This guide covers the beta-critical failures most likely to block LCopilot launch readiness.
Check:
/auth/mereturns the expected normalized role/onboarding/statusreturns expected company and onboarding details- the frontend is not still using a legacy auth context for the target dashboard
Symptom:
- exporter user lands in importer or another dashboard
- importer lands in exporter unexpectedly
- user sees a dashboard that does not match current onboarding truth
Check:
- local storage and session storage are actually cleared
- the surface is using the shared auth hook instead of a legacy context
- sidebar and dashboard read the same user source
Symptom:
- old user name appears after switching accounts
- UI shows mixed identity details
Check:
GET /api/results/{jobId}returnsstructured_resultstructured_result.versionisstructured_result_v1- frontend results mapping is rendering backend truth rather than fallback state
Symptom:
- overview counts differ from issues or documents
- verdict or readiness looks inconsistent
- key result surfaces disappear after refresh
Check:
- backend is reachable
- auth is valid
- uploaded files and form fields are present
POST /api/validateis receiving the correctuser_typeandworkflow_type
Useful endpoints:
/docs/healthz/health/live/health/ready
Check:
- the job id is valid
- the validation session exists
- persisted
structured_resultwas written to the session GET /api/results/{jobId}returns the stored payload
Use the beta-critical commands first:
npm run build
npm run test
cd apps/web && npm run test && npm run build
cd apps/api && pytestIf tests fail in old or non-beta surfaces, separate that from failures in the launch-critical exporter, importer, auth, or result-contract paths.