Important Information

01Built with OpenAI/GPT assistance

DesignsCtrl developed this tool with the assistance of OpenAI/GPT services. That describes how the tool was built; it does not mean each report is generated or certified by ChatGPT. The current reports come from programmed checks of fetched HTML and HTTP responses, plus available third-party measurements.

DesignsCtrl operates the tool independently. OpenAI and ChatGPT do not endorse or certify these reports. The operator, the code and any AI-assisted development can all make mistakes.

02Verify before changing your website

If you are unsure about a report, inspect the page and resource yourself, ask an experienced developer or qualified team member, or use another appropriate tool. You can share selected findings with an AI tool for a second opinion, but do not include private credentials or confidential data, and verify that opinion too.

Check the evidence, affected URL, source tag, time of scan and intended behavior. Back up relevant content before making changes. A source snippet or image thumbnail helps identify a resource; it is not a screenshot of the rendered section.

03How to read the statuses

Confirmed issue: the stated problem is supported by the response or source evidence inspected. Its significance and suggested fix still require context.

Needs review: the observation may be intentional or cannot be classified reliably from an automated check. Empty image alt text, cross-domain canonicals and script controls can be legitimate.

Could not check: access restrictions, timeouts, provider quotas, scan limits or unavailable capabilities prevented reliable verification. This is not proof that the site is broken.

Passed: the specified check completed without finding the particular problem. This does not certify the rest of the page or website.

04Scope, sampling and unavailable checks

A quick scan audits only the submitted public page. You may select up to four additional pages from the same origin, for a maximum of five. Pages are not chosen automatically. Fewer may be checked if requests fail. Source resource, link and sitemap checks use bounded samples and a total deadline.

HTML is fetched without running the website’s JavaScript. Dynamic content, rendered image candidates and browser-only behavior can be absent. Form fields, labels, validation, submission endpoints and form behavior are excluded.

A deployed browser-rendering service is not currently connected. Mobile/tablet/desktop overflow and overlap, rendered contrast/ARIA checks, complete focus behavior, browser console errors and rendered lazy/srcset image checks are not implemented. Their absence is explicitly reported. Full keyboard navigation needs manual review.

05Thumbnails and exported reports

Image findings show the discovered source URL and an available thumbnail. Broken, blocked, unsupported or oversized images may have no preview; a missing thumbnail is not additional evidence of an image-loading failure. Other elements can show their source tag or text snippet.

PDF and HTML exports include the report’s evidence and successfully retrieved thumbnails. Exports do not retrieve every page as a visual screenshot. Preview retrieval is bounded; source URLs remain available when a thumbnail cannot be embedded.

06Performance, privacy and optional help

Lighthouse measurements are attempted through Google PageSpeed Insights for the submitted page on mobile and desktop. Quota failures and unavailable data are clearly labeled. Counts in the report are findings, not an invented website-quality score.

Scans can cause requests to external websites and providers. Review the Privacy Policy before submitting a URL, and avoid confidential query values. Reports are not saved in an application report-history database.

Suggested fixes are freely visible. DesignsCtrl help links are optional, and no contact form is submitted automatically. Contact prefill is not enabled; copy an issue summary if you want to include it in an enquiry.