Will It Vibe?
115 deterministic checks, zero AI in the scoring path

You vibe-coded it.
Will it survive production?

Point Will It Vibe at a repo or drop a folder and get an honest 0 to 100 code health score from 115 fixed checks. Scans run in your browser, so your code never leaves it.

Free scan. No sign up. Nothing uploaded.

Sample report, a typical vibe-coded repo

82.8out of 100

Mostly vibes, a few things to clean up.

  • Security90
  • Code Quality84
  • Architecture76
  • Docs & UX85
  • Testing & CI60
  • Dependencies88

How it works

1

Point it at your code

Paste a GitHub repo URL, drop a project folder, or pick a zip. No sign up needed for a scan.

2

115 checks run in your browser

Fixed, inspectable rules: regex, AST, and file-presence checks. A scan takes seconds and your code never leaves your machine.

3

Share it or fix it

Save the report to a shareable link, or hand the findings to your AI assistant with copy-paste fix prompts and rerun until it vibes.

Six categories, weighted by what actually breaks

Each category starts at 100 and loses a fixed number of points for every rule that fires. The overall score is the weighted average.

Security

30%

Leaked keys, injection paths, disabled TLS checks, missing safety nets. Weighted highest because a committed secret matters more than a missing changelog.

25 rules

Code Quality & Syntax

20%

Leftover debug statements, empty catch blocks that swallow errors, thousand-line files nobody can review.

25 rules

Architecture & Best Practices

15%

Hardcoded config, committed node_modules, missing lockfiles, structure that will not survive a second deploy.

20 rules

Testing & CI

10%

Whether anything checks your code before it ships. Tests, CI, linters, formatters, type checking.

10 rules

Documentation, UX & Accessibility

15%

READMEs someone can actually follow, plus accessibility basics: alt text, labels, the viewport tag, focus styles.

20 rules

Dependencies & Hygiene

10%

Unpinned versions, deprecated packages, Dockerfiles running as root, five-megabyte binaries in git.

15 rules

The score is free. The full breakdown is $9.

Free, every scan

Your overall score, the verdict, all six category scores, and the top 3 findings in each category. Enough to know where you stand.

Full report, $9 one time

Every finding with file and line, fix guidance per rule, copy-paste AI fix prompts, a PDF, and a score badge. Paying never changes the score.

Pro, $12/mo

Every report unlocked, scan history, repo watching with rescan nudges, and side-by-side compare.

Fair questions

Is this AI?

No. All 115 checks are deterministic rules with fixed point values. Run the same code twice and you get exactly the same score. That is the point: a score that drifts between runs is not a score you can trust, and there is no model anywhere in the scoring path.

Is my code uploaded?

Dropped folders and zips are scanned entirely in your browser and never leave it. For GitHub repos, our server proxies the tarball from GitHub to your browser and does not keep the code. If you choose to save a scan, only the computed report is stored: scores, findings, file paths and line numbers. Never source code.

Can the score be gamed?

Yes, in the sense that it measures what is checkable. You could satisfy every check and still build the wrong product. The score tells you whether known-bad patterns and missing safety nets are present, and it makes no claim beyond that. The methodology page lists every rule and its exact point value.

More detail on the scoring math is on the methodology page.

Find out in about thirty seconds

The scan is free and runs entirely in your browser. Worst case, you learn something before your users do.