Part of the Free Vibe Coding Safety Tools kit.
LaunchCheck helps vibe coders test the thing they are about to ship.
You give it a ZIP file or an HTML app. It checks the build, makes a report, and writes a repair note you can paste into your next AI coding chat.
It is for this moment:
"My AI says the app is done. Is it really done?"
Use LaunchCheck after the build exists and before you share it.
Watch the short demo:
demo/recordings/launchcheck-demo.webm
Record it again:
npm run record:demoLaunchCheck looks for common ship blockers:
- missing files
- broken ZIP structure
- JavaScript syntax errors
- browser console errors
- failed requests
- mobile, tablet, and desktop layout problems
- missing text or buttons you said should exist
- weak release signs
- repeat problems between builds
It does not run random apps, installers, scripts, or EXE files from a download. It treats every build like it might be unsafe.
- Node.js 20 or newer
- PowerShell on Windows
- A ZIP file or HTML file you want to check
For full browser checks, install Chromium:
npm install
npx playwright install chromiumFor a first simple check, you can skip Chromium and use --no-browser.
Open PowerShell in the LaunchCheck folder.
Install the project:
npm installCheck a ZIP:
node .\bin\launchcheck.js validate "C:\Path\To\your-build.zip"Check one HTML file:
node .\bin\launchcheck.js validate "C:\Path\To\index.html"Run a simple static check only:
node .\bin\launchcheck.js validate "C:\Path\To\your-build.zip" --no-browserLaunchCheck prints something like this:
[LaunchCheck] PASS: index.html
[LaunchCheck] Score: 69/100
[LaunchCheck] Verdict: Acceptable for Human Review
[LaunchCheck] Report: C:\...\validator-report.html
[LaunchCheck] AI packet: C:\...\ai-iteration-packet.txt
Open the HTML report first:
validator-report.html
Paste this file into your next AI coding chat when you want fixes:
ai-iteration-packet.txt
That packet is the main handoff. It tells the AI what failed, what changed, and what to fix next.
Ready for Release
The build looks good enough to ship based on the checks.
Acceptable for Human Review
The build may be okay, but you should read the report before shipping.
Needs Repair
Do not ship yet. Give the AI packet to your coding agent and run LaunchCheck again after the fix.
Use LaunchCheck on the final build, not your messy source folder.
Good inputs:
- the ZIP you plan to send
- the
index.htmlfile for a one-page app - a fresh build dropped into Downloads
Bad inputs:
- your whole dev folder
node_modules- random installer files
- code you are not ready to test yet
Watch mode is useful when your AI tool keeps making new ZIP files.
node .\bin\launchcheck.js watch --watch "$HOME\Downloads"Now drop a new ZIP or HTML file into Downloads. LaunchCheck will test it when the copy is done.
On Windows, you can also run:
.\Start-LaunchCheck.ps1LaunchCheck ignores old files when watch mode starts. It checks new files.
LaunchCheck makes these folders beside the file it checked:
_VALIDATION_REPORTS_VALIDATION_STATE_VALIDATION_WORK_VALIDATED_BUILDS
Start in _VALIDATION_REPORTS.
Look for:
validator-report.htmlai-iteration-packet.txt
You can add a launchcheck.qa.json file to your build.
You do not need this for your first run.
Use it when you want checks like:
- the page must say "Export"
- the Start button must exist
- clicking a button must show new text
- a function must exist on
window
Templates are in templates.
launchcheck validate <zip-or-html>
launchcheck watch --watch <folder>
launchcheck status
Help:
node .\bin\launchcheck.js helpnpm test
npm run test:syntaxLaunchCheck opens HTML in a browser test and reads files from ZIPs. It does not auto-run random native programs.
The public release also does not self-update. If you download a newer LaunchCheck ZIP, it will be treated like any other build to inspect. It will not replace the copy you are running.
Still, only test builds you trust enough to inspect.
MIT. See LICENSE.