spec: regression testing / vibe coders

Ship.
Break.
Repeat?
Solved.

paste your url. get a testability score. fix it in cursor. ai browser tests — zero test code required.

ref: health-check-v2.1
app health checkfig. 01
00
testability score
button labels
✓ resolved
duplicate names
✓ resolved
dynamic ids
aria labels
5 critical
→ fix prompt readycopy →
// works with
procedure / three-step process

Zero engineers.
Three steps.

no setup, no pipelines, no test files. paste a url and let the agent do the work.

step 01
Paste your URL
no install. no npm. no config. drop in your app's url — localhost works too. we open a real browser and start scanning.
→ any web app · localhost supported
step 02
Get your score
we surface every issue blocking reliable tests — missing labels, duplicate names, weak selectors. scored out of 100.
→ visual overlay · full report
step 03
Fix it, test it
copy the ai-generated fix prompt into cursor. re-scan to confirm. describe tests in plain english and run on every deploy.
→ ai tests · real browser · no code
component: health check

Your app has a
testability score.

before you can test anything, your app needs to be testable. most vibe-coded apps aren't — and you'd never know until a test silently fails.

  • scans for missing aria labels, duplicate names, weak selectors
  • visual issue overlay — see exactly where problems are
  • scored out of 100 — track improvement over time
  • free forever
health-check.vta
https://myapp.lovable.appscan →
71
testability score
buttons with no label
3 critical
dynamic ids — fragile
form labels correct
✓ pass
component: fix prompts

Copy. Paste.
Ship the fix.

every issue generates a plain-english fix prompt you paste straight into cursor, lovable, or bolt. no debugging. no stack traces.

  • no code knowledge needed — just paste and run
  • export all prompts at once or one at a time
  • re-scan to confirm the score improved
fix-prompt-01.vta
generated fix prompt
the "submit" button on your checkout form is missing an accessible label. add aria-label="submit order" so testing agents and screen readers can identify it reliably.
→ paste into cursorcopy →
component: ai testing

Describe the test.
We run it.

write tests in plain english. vibetestagent controls a real browser — clicking, typing, asserting — exactly like a human would, on every deploy.

  • no selectors, no code, no fragile css paths
  • video recording of every run
  • plain english failure explanations
  • group into suites, run after every deploy
test-run / checkout-flow
checkout_completerunning
navigate to checkout page341ms
fill in shipping address1.2s
click "continue to payment"890ms
payment form should be visibletimeout
generating failure explanation
3passed
1failed
1running
pricing / tier specification

One more tool
in your stack.

same price as cursor. no per-seat nonsense. start free.

free
$0/mo

start here. scan, score, generate fix prompts.

  • 3 health check scans/month
  • 5 ai test generations
  • 1 saved suite (5 tests)
  • fix prompt generation
start free
pro
$19/mo

for builders shipping daily. unlimited scans, real regression testing.

  • unlimited health checks
  • 50 ai test generations
  • unlimited suites + tests
  • video recording on all runs
  • ai autofix (20 retries/mo)
get pro →
team
$49/mo

multiple apps, scheduled runs, ci/cd when you're ready.

  • everything in pro
  • unlimited ai generations
  • multiple app profiles
  • scheduled runs daily/weekly
  • github actions integration
get team

annual billing · 20% off · cancel anytime

initiate scan

Your score is probably
under 50.

find out in 60 seconds. no signup, no install, no credit card.

// no signup · no install · no credit card