LaunchVet

Is the app you built with AI actually ready to ship?

LaunchVet reads the public page and gives you a letter grade and the top issues in plain English. The Launch Report adds every finding and the fix steps for Lovable, Bolt, Replit, v0 or Cursor.

Free scan shows the grade and the top 3 issues. No card.See pricing

Report/All issuesExample data
Grade F, hold the launch42 out of 100 · 7 issues · 6 areas read
100 − 5842 / 100
Critical 1
Medium 3
Low 3
LV-1

Stripe secret key in the JavaScript.

Severity
Critical
Area
Secrets in the bundle
Points off
−28

A live secret key is in a file the browser can download. Anyone can copy it and create charges or refunds on the account. We did not use the key.

Fix prompt

Remove the Stripe secret key from the JavaScript and from source maps. Rotate it in the Stripe Dashboard. Keep only the publishable key in the browser.

The report above is a sample with example data. Select an issue to read it. The free scan shows the grade, the score, the issue counts and the top three issues. The full table, including what passed, is in the report.

Sample report

A grade, then the reasons. Every issue is written in plain English, with a prompt you can paste into your builder.

  • Every failed check costs points

    Each check that fails takes points off a score of 100. The score becomes the letter.

  • The free scan shows the top

    The grade, the score, the issue counts and the top three issues. The full table, including what passed, is in the report.

  • One rule to know

    A single critical issue, such as a live secret key or a site stuck on HTTP, caps the grade at F. Passing checks do not cancel a leaked key.

How the sample adds upExample data
  • Starting score100
  • Secrets in the bundle1 issue−28
  • Headers and HTTPS2 issues−14
  • Supabase and FirebaseNoticed, never queried0
  • Search and sharing2 issues−6
  • Weight and speed1 issue−3
  • Privacy and terms1 issue−7
  • Score42
  1. AReady
  2. BClose
  3. CNeeds work
  4. DNot yet
  5. FHold the launch

How it works

Three steps, and you touch nothing. Paste a URL, say it is yours, and read what comes back.

  1. Step 1

    Paste the public URL

    Tick the box to confirm you own the app, or you’re allowed to test it. We won’t scan a site you don’t name.

  2. Step 2

    We read what a visitor can read

    The HTML, same-origin scripts, headers, robots.txt and a sitemap. No login, no writes, no database queries.

  3. Step 3

    You get a grade and the top 3

    Each issue is in plain English, with a fix prompt you can paste into Lovable, Bolt, Replit, v0 or Cursor. Leave an email and the rest of the list opens.

What we check

Six areas of the public page. Nothing behind a login.

Passive and read-only. Active checks, such as asking a Supabase table whether it will return rows, wait until you have proved you own the app. They are not in this version.

Bundle

Secrets in the bundle

Live keys, private keys and source maps in the JavaScript a browser can download. Values are redacted.

Example hint. The start of the key, its last four characters and the file name are kept so you can find it. The rest is not stored.

Transport

Headers and HTTPS

HSTS, content security policy, clickjacking, the certificate, and whether HTTP is still open.

Data config

Supabase and Firebase

We notice a public config. We do not connect, and we do not read a single row.

Search

Search and sharing

Title, description, Open Graph, language, robots.txt and a sitemap.

Performance

Weight and speed

A hint about response time, HTML size, script size and compression. Not a lab benchmark.

Legal

Privacy and terms

Whether the public page actually links to them. The wording still needs a lawyer.

Every example on this page comes from the same sample report. None of it is a live result.

Why this exists

The pattern is well documented. A 2025 report on Lovable apps found Supabase tables that answered anyone who asked.

The Lovable position is that each builder is responsible for their own app. That is the gap LaunchVet sits in, and it is not only a Lovable problem.

Reference: CVE-2025-48757

Reads

  • The public HTML
  • Same-origin scripts
  • Response headers
  • robots.txt
  • A sitemap

Never

  • Logs in
  • Writes anything
  • Queries a database
  • Reads a row
  • Scans a site you did not name

Pricing in USD

A free checkup, then a one-time report if you want the lot.

The free scan is the grade and the top issues. The Launch Report is $29 once for that scan. Watch and the other plans are coming soon.

Free

Free scan

$0

No card

A letter grade, a score, and a summary of the top issues.

  • One public URL
  • Grade, score and issue counts
  • A summary of the top issues
  • Passive checks only

One-time

Launch Report

$29

One-time

Every finding for this scan, with the fix steps.

  • Every finding, in plain English
  • Fix steps for your builder
  • Checks that passed
  • Tied to one scan

MonthlyComing soon

Watch

$13

Per month

A weekly look, so a new secret doesn't sit there until launch day.

  • Weekly re-scans
  • Alerts for a new secret or an open table
  • Up to 3 apps

Coming soon

One-timeComing soon

Fix-it-for-me

$165

One-time

We open a pull request with the fixes. A person reviews it.

  • An agent opens a pull request
  • A human reviews high-risk changes
  • Re-scan included

Coming soon

MonthlyComing soon

Agency

$65

Per month

For freelancers and studios shipping other people's apps.

  • Up to 25 apps
  • White-label PDF reports
  • The same checks as Launch Report

Coming soon

Prices are in US dollars.

Includes GST for Australian customers.

Watch, Fix-it-for-me and Agency are coming soon.

Questions

Straight answers.

Six of the ten questions on the FAQ page. Quicker than reading about it: the free scan takes one URL and no card.

Run a free scan
What do you actually check?

The public page only. We look for secret keys in the JavaScript and source maps, security headers, HTTPS and the certificate, Supabase or Firebase config sitting in the browser, title and social tags, robots.txt and a sitemap, how heavy and how slow the page is, and whether a privacy policy and terms are linked. We do not log in, and we do not try to break in.

Do you open my database or read customer data?

No. If we see a Supabase or Firebase config, we record that it is there and stop. We never send the anon key, the service role key, or a Firebase API key to those services, and we never read rows. A public anon key is normal. An open table is a problem, and checking that properly comes later, only after you prove you own the app.

Is this a penetration test or a certificate?

Neither. It is a launch checkup: a letter grade and a short list of fixes. It will miss things. It is not a warranty that the app is safe, and it is not legal advice about the Australian Privacy Act.

Why do I have to tick a box?

You are asking us to fetch a site and read its public files. That is fine for an app you own, or one you are allowed to test. It is not fine for someone else's. The box is how you tell us which one this is. We rate-limit the checkup as well.

What happens to secrets you find?

We keep a redacted hint, such as the first and last four characters, plus the file name, so you can find it. The full value is not stored in the report. Rotate the key anyway. A free scan is deleted after 30 days. A paid Launch Report is kept for 12 months.

What do the grades mean?

A is ready, B is close, C needs work, D is not yet, and F means hold the launch. A single critical issue, such as a live secret key or a site stuck on HTTP, caps the grade at F. Passing checks do not cancel a leaked key.

Paste a URL. Get a grade.