The diagnostic, and its limits

The diagnostic reads a public website the way a visitor would. It tests no vulnerability, submits no form, logs into nothing.

  1. You paste your website address and run the diagnostic. No account, no email requested.
  2. A queue shows your position.
  3. Our crawler reads up to 25 pages of the site, accepting nothing and running no script, like a first-time visitor.
  4. The report shows five findings, each with its evidence: third parties before consent, accessibility, languages, internal linking, readability by AI assistants.
  5. The report has a private, shareable, non-indexed address.

What an automated test cannot see. Requests triggered by scripts once they run, the actual reading order, whether alternative text is meaningful, whether a form is clear. The report says so for each finding.

Usage limits

LimitValueReason
Diagnostics per IP address per day3Cost, abuse
Diagnostics per target domain per hour1 (the existing report is shown again)Not to serve as an attack relay
Pages read per diagnostic25Cost, courtesy
Duration per diagnostic2 minutes, then a partial report labelled as suchCourtesy
Queue5 simultaneous diagnostics; beyond that, your position is shownLoad
Refused targetsInternal addresses; domains on the exclusion list (public administrations, banks, health); domains whose robots.txt disallows our crawlerSecurity, respect
Crawler identificationUser agent OlvraSitesDiagnostic, with a link to this pageTransparency

The proof of work planned in our specification asks your browser to compute on our behalf: this site's modules only add interaction and motion, never a computation for us. It is replaced by the limits above, enforced on our servers.

What we keep

  • The report: 90 days, then its content is erased. It only contains what any visitor can read on the site diagnosed.
  • The rebuilt preview, when there is one: 7 days. Private to its link, not indexed, never published or reused.
  • Your IP address: never written to a database. A fingerprint of its truncated prefix is used, in memory and for the day only, to enforce the three-diagnostic limit.
  • No email, no account, no cookie.

Opting your site out of our crawler

Our crawler identifies itself as OlvraSitesDiagnostic. To keep it from reading your site, add two lines to your robots.txt file: “User-agent: OlvraSitesDiagnostic” then “Disallow: /”. The diagnostic is then refused before any reading.

For any other request: hello@elahadvisory.ai.

Run the diagnostic

No personal data requested. What we keep, and for how long.