Model Context Protocol

Give your AI a security scanner.

Connect SafetyVibe to Claude, Cursor, Lovable, Replit, VS Code and more. Then just ask your AI to check a site — it scans security, privacy, performance, SEO and what AI tools can see, and tells you what to fix.

Server https://mcp.safetyvibe.eu/mcp Remote · Streamable HTTP · no API key, no login required.

What your AI can do

scan_site

Scan any public website from the outside — security headers, HTTPS/TLS, exposed files & secrets, known-vulnerable libraries, privacy & trackers, SEO and whether AI crawlers can read it, plus performance & accessibility scores when the page can be rendered. Plain-language findings, each with a fix.

run_access_tests

Test your Supabase Row-Level Security with two throwaway logins: can one signed-in user read or write another user's data? Consent-first, your own project only — use disposable test accounts, never a real password.

get_security_rules

Concrete security rules to follow before you build a feature (e.g. Supabase auth, payments, storage).

define_access_model

Turn a plain-language access model ("customers see only their own orders") into a structured spec you can verify.

Add it to your tool

How to use it

Once it's connected, just talk to your AI in plain language — it picks the right tool automatically:

you › Scan example.com and tell me what needs fixing

you › Check my site's security and SEO before I launch: mysite.com

you › Is my Supabase row-level security safe? Here are two test logins…

you › Can AI crawlers and ChatGPT read my site? Scan mysite.com

A full scan runs several checks and can take a minute or two — your AI will wait and then summarize.