Lighthouse is how you find out whether the thing you built is actually any good. It loads your page and scores it across four axes - performance, accessibility, best practices, SEO - returning specific, fixable findings instead of a shrug. Built into Chrome DevTools, and scriptable via CLI and CI.
Where it shines
- The feedback loop AI builds need. Generated code is fast to produce and slow to judge; Lighthouse turns judgment into a number you can diff after every change.
- Accessibility as a checklist, not a lecture. Contrast failures, missing labels and broken landmarks come with selectors and links to fixes.
- CI-native. Run it in pipelines and block merges when the performance budget regresses - quality enforcement without a QA team.
Where it falls short
- Lab scores simulate a device and network; they don’t capture what your real users experience (field data).
- A perfect score doesn’t prove UX quality - it proves the tested criteria pass.
The real cost
Free and open source. It costs you the honesty to act on the report instead of closing the tab.
Verdict
The single highest-value habit for vibe coding: after the agent finishes a feature, run the audit before you celebrate. Numbers beat vibes.
