Website security and AI readability audit, tested on oats.lv

I built two passive website scanners and ran them on my own company site first. Then I fixed what they found and scanned again.

The security scan runs 17 checks from the outside: security headers, HTTPS and the certificate, email authentication (SPF, DKIM, DMARC, MTA-STS), exposed files, outdated software and signs of a hacked site. The AI readability scan runs 8 checks on whether ChatGPT, Perplexity and Google's AI answers can load the homepage, read its text without JavaScript and find the company data, sitemap and llms.txt. Neither scan logs in, installs anything or changes the site.

Results on oats.lv, 9 November 2026. Security score, out of 1000: 902 before the fixes, 926 after, grade A. Security findings: 9 before, 5 after. AI readability score, out of 100: 76 before, 90 after. AI readability problems: 3 before, 1 after.

Every finding in the report says what it means, who can change it and what the fix costs. Fixes start at 49 EUR, and the total never passes 499 EUR however many a site needs. The app runs on Python and Flask in Docker on a Hetzner server, behind a signed proxy from the Netlify site.