Skip to content
Tomas KadlecBook a Call

Is your AI-built website safe to launch?

Sites built with AI tools often ship with exposed API keys, open config files, missing protections and forms anyone can abuse.

Enter a page. You get the problems a stranger could find from the outside, in plain language, with a fix you can paste into your AI builder.

Check a website

Without an e-mail you get the short result on screen and nothing is stored beyond an abuse log. With an e-mail I send the full report with fixes, keep the e-mail, the address and the source IP for 12 months and may write to you once about the result. Details in the Privacy.

What it checks

  • Secret keys in the code

    Privileged keys (Stripe, OpenAI, Anthropic, Supabase service-role, AWS) left in the page or its JavaScript. Public keys are not flagged.

  • Exposed files

    Publicly readable .env, .git, backups or config files. The content is verified, not just the status code, so a single-page app does not cause false alarms.

  • Transport and headers

    HTTPS, the HTTP to HTTPS redirect, mixed content, and the security headers (CSP, HSTS, clickjacking, and more).

  • Forms and cookies

    Forms that submit over HTTP or to another domain, sensitive fields in the URL, missing CSRF protection, and session cookie flags.

  • CORS, source maps and libraries

    An API that allows any origin, published source maps, and outdated libraries with known vulnerabilities.

  • Active test on your own domain

    Verify you own the domain and the tool probes your forms for reflected XSS and SQL injection, for detection only.

Active tests run only on your own domain

The passive check works on any URL and never attacks the site. The active test, which sends harmless test input into forms to find XSS and SQL injection, runs only after you prove you own the domain with a DNS record or a file. That keeps the tool from being pointed at other people's sites, and keeps you on the right side of the law.

What it cannot tell you

From the outside this cannot test your database rules, who can log in, payment logic or private API routes. It reads one page and a short list of your site's own paths, so it does not crawl thousands of pages or overload anything. A clean result in what was tested is not a promise that the whole app is safe.

Will it attack my site or other sites?

The passive check never sends attack input, to any site. The active test sends only harmless, non-destructive detection probes, and only to a domain you have verified you own. It never reads or changes your data, guesses passwords or uses a key it finds.

How do I run the active test?

Run the passive check first. The result page shows a DNS record and a file, either of which proves you own the domain. Add one, click the button, and the active test runs. The value stays the same, so you add it once.

Which page should I check?

Start with the home page, then a page with a form or a logged-in area. Different pages load different code.

Does it work for larger sites?

Yes. It checks one page plus a fixed set of your domain's own paths, so the load is the same whether your site has five pages or fifty thousand.

Found problems you want fixed?

I fix the blocking problems, give you the prompts for the rest, and run the check again to prove it is gone.

Book a free 20-minute callSee all free tools