Execute declared checks
Detect lint, typecheck, test, and build commands. Keep exit codes, timing, and bounded output in the report.
Release evidence, not release theater
Know what blocks release before users do.
ShipVitals audits AI-built apps before they are published, sold, submitted, deployed, or delivered. It runs what can be proved, records what cannot, and returns one release verdict.
$ npx github:omarkhandji-commits/shipvitals audit .20Open-source audits
4Evidence verdicts
6Proof levels
0Telemetry by default
A gate, not another scanner.
Specialist tools produce signals. ShipVitals turns those signals, missing proof, and product context into a release decision that another person can inspect and rerun.
Look for false completion
Flag secret candidates, skipped tests, mock data, demo-only paths, and missing product promises without treating every match as a blocker.
Cap unsupported confidence
Missing runtime, visual, CI, or independent evidence limits the score. No proof means not verified, never passed.
Different question. Clear boundary.
ShipVitals does not compete with the tools that inspect a narrow technical domain. It asks whether their evidence is sufficient for the next release destination.
| Tool | Primary answer | ShipVitals adds |
|---|---|---|
| Tests | Does known behavior pass? | Promise, missing proof, release verdict |
| Lighthouse | How does a page perform? | Auth, payment, support, and launch context |
| Playwright | Does a browser flow execute? | Evidence level and confidence cap |
| Security scan | Are known risks present? | Product, UX, delivery, and acceptance context |
| Code review | Is the implementation sound? | A reproducible go/no-go decision |
Run before release
Make the decision inspectable.
Start with static and deterministic evidence. Add runtime, visual, CI, and independent proof only when someone actually observed it.