About Crash Dummy
What exactly is Crash Dummy?
Crash Dummy is the crash-test stage for apps built with AI tools like Lovable, Bolt, Replit, Cursor, v0 and Base44. We test your live app from the outside. First a stranger with no account tries it. Then, in the Full Crash Test, two test users try to get at each other's data. You get a Crash Report with a 1–5 Safety Rating, plain-English findings, and repair prompts you paste back into your AI builder.
Do I need to be a developer?
No. The report is written for founders, designers and makers, not security engineers. Every finding explains what happened and why it matters. The fix comes as a prompt you can paste into the tool you built with.
Which builders and stacks do you support?
Today we're strongest on apps that use Supabase, which includes Lovable Cloud (it's built on Supabase). The two-dummies test works from the outside, so in principle it works on any web app with user accounts. Using another stack? Email us before ordering, and we'll tell you honestly what we can and can't cover.
What's the difference between a Dummy Run and a Full Crash Test?
A Dummy Run ($49) sends in the Stranger: no account, no password. It checks what anyone on the internet can see: secrets in your public code, tables readable without login, public files, exposed functions, and basic web hygiene. A Full Crash Test ($149 founding price) adds the two-dummies test: can one user read, change or delete another user's data, or give themselves admin or the paid plan? It also covers sign-up and sign-in, payment bypass (Stripe test mode only) and your critical flows on mobile and desktop. A human tester reviews every finding, and one free Re-crash is included.
What is Airbag?
Airbag ($29/month) is coming soon. It will re-run your crash test on a schedule or after each deploy, and alert you when your Safety Rating drops. You can join the list now. You won't be charged until it exists.
Compared with other tools
Lovable (or my builder) already has a security scan. Do I still need this?
Keep using it. It's a good first line. Lovable's Quick scan, for example, runs when you publish and checks for tables without row-level security, rules that let everyone through, and vulnerable dependencies. Lovable's own docs say these tools don't replace a thorough security review. A scan reads the blueprint. A crash test drives the car into the wall: we check what a second user can actually reach in your live app, and show you a replay of it. Run the scan, fix what it finds, then crash-test.
Is this a penetration test?
No. A crash test is focused on the problems AI-built apps hit most often: data one user can see from another, self-upgrades, leaked secrets, public files, abuse of email features, and broken critical flows. It doesn't replace a full penetration test, and we don't run fuzzing, load tests or attacks on your infrastructure.
Safety and privacy
Will this break my app or touch my users' data?
We're deliberately gentle. We send at most 5 requests per second, and abuse checks stop at 10 requests. We never run load tests. In a Full Crash Test, all writes happen in the two test accounts you create, and we clean them up afterwards. When something is exposed, we record that it's exposed and how many rows, never the contents.
Can I crash-test someone else's app?
No. We only test apps whose owner has verified ownership, with a tag on the site, a file, or a DNS record. Unverified addresses are refused. That includes competitors, clients who haven't agreed, and "just curious."
Do you need my code or my builder login?
No. We test the live app from the outside, the way your users reach it. We never ask for your Lovable, Bolt, Replit or Supabase password.
What do you keep, and for how long?
We keep the evidence that supports each finding, with other users' data redacted and secrets masked. Evidence is deleted 30 days after your report. Your report is private to you.
What happens if you find something critical?
In a Full Crash Test, we tell you right away and don't wait for the report. You can fix it immediately and ask for a Re-crash.
Results
Does a 5-star Safety Rating mean my app is secure?
No. No test can guarantee that. Five stars means our dummies found nothing confirmed in the areas we test, on the date of the test. Your next AI edit can change that, which is why every report is dated.
How is the Safety Rating calculated?
With a fixed, published rule. Every app starts at 5 stars. Any Totaled finding caps it at 1, any Crash at 2, two or more Dents at 3, and one Dent at 4. Each Scratch then removes half a star, Scratches together remove at most one star, and the rating never drops below 1. Only confirmed findings count. The full rule is on the methodology page.
What is a repair prompt?
It's a fix written as an instruction for your AI builder. For example: "Turn on Row Level Security for the tasks table and only let each user access their own rows." You paste it into Lovable, Cursor or whichever tool you used, then press Re-crash to confirm it worked.
What is a Re-crash?
After you apply the fixes, we run the same tests again and update your Safety Rating. One Re-crash within 30 days is included in the Full Crash Test.
Ordering
How fast do I get my report?
A Dummy Run report arrives by email within 24 hours. A Full Crash Test is delivered within 72 hours of your signed testing authorization. If we miss the deadline, you get a full refund.
What do I need to prepare?
For a Dummy Run: your app's URL, one minute to add the verification tag, and accepting our testing terms when you order. For a Full Crash Test: also two test accounts, created through your app's normal sign-up, and the one-page authorization, which we email you after you pay.
Can my AI agent apply the fixes?
Yes, and that's the idea. Every report comes with a fix plan written for your agent: what to fix, in what order, how to check each fix, and what never to touch. Click "Hand this to your agent" (or save the file to your project), and tell your agent to apply it. Then press Re-crash to confirm. You don't need to understand the fix yourself.
Can my AI agent run the crash test itself?
Not yet. An MCP server (for Lovable, Cursor, Claude Code, Bolt and Replit), a GitHub Action and a CLI are in early access, and none of them is live. What works today is a one-time snippet for your agent's instructions (AGENTS.md, CLAUDE.md, Lovable Instructions or Cursor rules). With it, your agent reminds you before risky deploys and knows how to apply a fix plan safely. Your agent will never pay for anything. You always decide.
Who's behind Crash Dummy?
Ron Rosenzwieg, a QA and test-automation engineer who works with Playwright, API and end-to-end testing, and testing of AI and MCP systems, with a background in security. Crash Dummy is new. Founding clients get a lower price in exchange for honest feedback.