Incident Response

May 30, 2026 | 5 min read

Detection

  1. Monitor fallback rate spikes in the dashboard.
  2. Set up alerts for correctness_match=False events.
  3. Watch for compile credit exhaustion.

Triage

pyvorin doctor --network --full
pyvorin job details JOB_ID
pyvorin report

Mitigation

  1. Enable force fallback: PYVORIN_FORCE_FALLBACK=1
  2. Delete corrupted cache: rm -rf ~/.pyvorin/cache/*
  3. Rollback to previous artifact version.

Post-Incident

  1. Capture failure event JSON.
  2. Run pyvorin report-failure to send diagnostics.
  3. Document the fix in runbooks.