Storage and recovery paths, not SQL or your ORM, are where SQLite surprises…

Storage and recovery paths, not SQL or your ORM, are where SQLite surprises show up; the recent 16-year WAL-reset bug Tailscale helped expose proves that. If you rely on SQLite for agent state, caches, edge services, or embedded workflows, add tests that exercise process restarts, interrupted checkpoints, and filesystem faults, not just query correctness. Run integrity_check against backups and replay write logs under failure injection to catch invisible lost-commits before they hit production. The rule I’d use: if you can restart the process or corrupt a page in a lab, write a test for it.