Awlware Link Checker

Description

Awlware Link Checker finds links and images that are really broken and helps you fix them where they are used. Everything runs on your own server: there is no account, no cloud service and no limit on the number of links.

It tells “broken” apart from “the site refuses robots”. Social networks, marketplaces and job boards often answer automated requests with 403, 429, 999 or a captcha page while the page works fine in a browser. Those links are shown as Blocked (verify manually) instead of cluttering the Broken list. Server errors (5xx) and timeouts are rechecked later and reported as broken only if they repeat.

It looks where links actually are: posts, pages and public post types (content and excerpts), classic menus, block theme navigation, synced patterns and templates, custom fields you choose, Text, Custom HTML and Block widgets, pages built with Elementor, and (optionally) approved comments and commenters’ websites. Images (including srcset), iframes, video and audio sources are checked too.

Fix links from the report: Edit URL or Unlink in one or all places (menu items and navigation links stay as plain text), Remove image, Recheck, Dismiss. Posts keep a revision, every change is written to a log and can be undone. Elementor layouts are shown with a link to the Elementor editor.

Light on your site: checks run in short background slices (WP-Cron, 20 seconds every 5 minutes), one request at a time per external site with a pause between requests. Links to your own content are verified from the database instead of over HTTP. Nothing runs on visitor page views. Database tables are created and updated only through dbDelta under a lock, never on visitor requests.

How a link is judged

• Broken: 404 or 410, domain does not exist, redirect loop, missing anchor on one of your pages, or a server error / refused connection that repeated on separate checks.
• Unconfirmed: a server error or refused connection on the first check. It is checked again after an hour (then 2, 4… hours) and becomes Broken only if it fails again.
• Timeout: the server did not answer on repeated checks.
• Suspected: the page answers “200 OK” but looks like a “not found” page (soft 404): its title or text says so, WordPress shows its 404 template, or a deep link redirects to the home page. Never counted as broken.
• Blocked: anti-bot protection (Cloudflare, PerimeterX, DataDome, Akamai, Imperva, Sucuri, AWS WAF, captcha pages), rate limiting (429), login required (401), or a site known to refuse automated checks.
• Unknown: the answer does not prove anything (for example 403 without anti-bot signs or a TLS error).
• Redirect and OK: the link works.

WP-CLI

wp awlware-lc scan, `index`, `report --format=table|json|jsonl`, `where <url>`, `stats`, `reset`.

For developers

Actions and filters start with awlwlc_, for example awlwlc_user_agent, awlwlc_request_args, awlwlc_post_types, awlwlc_meta_keys, awlwlc_exclude_url, awlwlc_classify_result, awlwlc_verdict_changed, awlwlc_results_saved, awlwlc_report_tablenav, awlwlc_admin_tabs.

External requests

Checking links means requesting them, so this plugin makes outgoing HTTP requests. Here is exactly what it does:

• It requests only the URLs found in your own content (HEAD first, GET when needed) to see whether they work. Nothing else is fetched.
• No data is sent to the plugin author or to any third-party service. There are no Awlware servers involved, no telemetry, no license checks and no remote lists: the list of anti-bot sites ships with the plugin.
• Each request carries the User-Agent Mozilla/5.0 (compatible; AwlwareLinkChecker/1.0.0; +https://awlware.com/link-checker/), so site owners can see who is checking. It can be changed with the awlwlc_user_agent filter. The plugin never pretends to be a browser.
• One request at a time per external site, with a pause between requests, and at most three attempts per URL per check (redirects are followed hop by hop). Soft-404 detection on other sites (one extra GET per working page) is off unless you enable it. A site that does not resolve or refuses connections is not asked again in the same run.
• All requests go through the WordPress HTTP API, so proxies, WP_HTTP_BLOCK_EXTERNAL, WP_ACCESSIBLE_HOSTS and HTTP filters apply.
• SSRF protection: addresses in private, loopback, link-local and reserved ranges (IPv4 and IPv6) are never requested, every redirect is checked again, and the checked address is pinned for the connection to prevent DNS rebinding.

Privacy

The plugin stores the list of links, where they are used, their check results and a log of fixes (who changed what, kept 90 days) in its own database tables on your site. No data leaves your site except the link-check requests described above. It does not use cookies and adds nothing to the public side of your site.

Technical details