Website Health Scanner changelog

November 1, 2026: clearer mobile findings

• Report an omitted device-width hint as a signal to check, rather than proof that the mobile layout is broken.
• Show example page URLs for viewport, table, and fixed-width signals so readers can inspect the affected pages.
• Publish a dated self-scan and explain how the four reported tables were located and what remains to be tested.
• Treat an absent HTML canonical link as context instead of an automatic SEO failure, with a dated real-world example and a note about unchecked HTTP headers.

These changes make a report easier to interpret. They do not certify a scanned site or establish AdSense eligibility.

October 4, 2026: compact reports and fresh repeat scans

• Show the three highest-priority findings first, with remaining checks and measurements in expandable sections. Copy, JSON export, and printing retain the full report.
• Display the scan timestamp and distinguish a cached result from a new scan.
• Add a guarded fresh scan action that bypasses both cache layers. Existing workload limits remain in effect, with an additional refresh cooldown and quota.
• Keep the previous report available when a repeat scan fails, and show the retry delay when a fresh scan is temporarily limited.

October 3, 2026: fewer false positives in reports

• Blocked recognized sensitive URL parameters, kept query URLs out of shared caches, PageSpeed, and browser history, and disabled detailed outbound-request traces.
• Bounded in-memory scan caches and rate-limit records, tightened API origin and input checks, and updated dependencies with security fixes.
• Validated Node DNS results at socket connection time, redacted blocked discovered URLs, handled invalid base elements, and rejected incomplete Lighthouse measurements.
• Moved the repair plan and severity-sorted findings ahead of the detailed measurement panels without removing any report data.
• Kept successful reports visible when browser history storage is blocked or full, ignored invalid saved entries, and prevented history clicks from changing the URL during a scan.
• Stopped treating omitted or empty form actions as broken. Recognize external form-associated controls and exclude disabled fields and non-data buttons from unnamed-field warnings.
• Added affected page URLs and an explicit reminder that form submission and delivery are not tested.
• Removed arbitrary title and description length grading and excluded intentional noindex pages from search metadata checks.
• Added regression fixtures and versioned the scan cache so previously cached rule results are not reused.

The earlier self-audit snapshots remain visible with a correction explaining why their scores can differ from a new run. A change in scoring rules is not a repair to the target website or a claim of advertising approval.

September 4, 2026: evidence and editorial accountability

• Published a dated self-audit using real production scan values and manual mobile checks.
• Added visible authorship and review dates to the site's technical guides.
• Expanded the operator page with the maintenance, verification, correction, and disclosure process.
• Removed privacy, terms, and imprint pages from the XML sitemap and marked them noindex,follow. They remain accessible to people and crawlers through the footer.
• Kept advertising delivery disabled while preserving the AdSense ownership tag and ads.txt entry needed for review.

The release was validated with the automated application suite, a Cloudflare dry run, production endpoint checks, crawler user agents, responsive browser checks, and a live scan. The goal was not to create more URLs. It was to make the indexable part of the site demonstrably useful, attributable, and honest about its limits.

August 22, 2026: self-scan and mobile reliability

• Fixed the Cloudflare self-scan path that previously returned HTTP 522 when the Worker requested its own public domain.
• Enabled Cloudflare's public-network fetch behavior for the scan runtime while retaining private and local destination blocking.
• Checked 50 viewport combinations between 320 and 768 pixels and corrected page-level mobile overflow.
• Verified that unusually long target domains and dynamic report text wrap inside their containers.
• Improved scan progress feedback so a visitor can see the current phase and elapsed time during a longer crawl.

The earlier 522 was a deployment-context problem, not evidence that the public site itself was offline. Reproducing the request through the production Worker identified the difference between a normal external scan and a request back to the same Cloudflare origin.

July 24, 2026: content review and advertising safeguards

• Published practical guides for website repair, Core Web Vitals, manual audits, and contact-form testing.
• Added a public About page with operator, contact, funding, and editorial information.
• Disabled Auto Ads and all ad slots during the AdSense content review.
• Retained ownership verification without loading the AdSense advertising script on content, legal, loading, error, or report states.
• Added production checks that fail if advertising is accidentally enabled while the site remains in review mode.

This separation matters because ownership verification and ad delivery are different controls. A publisher can prove control of a site without placing ads on pages that have not yet passed review.

July 10, 2026: bounded crawler and production foundation

• Introduced a same-site crawl with explicit limits for pages, depth, links, images, redirects, time, and response size.
• Blocked private, loopback, local, and restricted network targets before public scanning.
• Added eight report categories covering speed, images, forms, buttons, mobile signals, broken links, search metadata, and Core Web Vitals.
• Added canonical URLs, permanent redirects from legacy HTML paths, robots.txt, sitemap.xml, security headers, and a JSON health endpoint.
• Deployed the application to a Cloudflare Worker behind the webhealthscan.com origin.

The crawler was designed as a diagnostic sample from the beginning. It does not attempt to mirror a whole website or send real form submissions. These boundaries reduce traffic to scanned sites and make the limitations of each report understandable.

How releases are checked

Each material release starts with targeted application tests, then a full project check. Cloudflare configuration is validated with a dry-run bundle before deployment. After deployment, the public health endpoint, canonical redirects, legal configuration, ads.txt, ownership metadata, editorial routes, sitemap, security.txt, 404 behavior, and scan API are checked against the production origin.

Browser checks are added when a change affects layout or interaction. Scanner changes are tested against both an external public site and webhealthscan.com itself. A release can still contain defects, so reproducible reports are reviewed and corrected rather than hidden. See the correction process or email the operator using the contact information on the About page.