SiteVerdict — Broken Link Checker

Find every broken link, image and script on your site, and see the exact page it sits on and the anchor text to fix. Internal and external links are checked once each, politely, with per-host rate limits. You also get redirect chains and loops, mixed content and missing #anchors.

Who it is for: SEO agencies and site owners doing link hygiene, content teams after a migration, developers who want a pre-release check, and AI agents (MCP-ready).

Three ways people use it

1. Monthly link audit. Crawl the whole site, then download ISSUES-CSV (one row per broken occurrence: source page, target, anchor text, status) and hand it to the editors.
2. Migration QA. After moving a site, find internal links that now redirect, redirect chains and 404s from old URLs.
3. Scheduled monitor. Run weekly with monitoring on (the default). A Slack/webhook alert fires only for new broken links or new active mixed content.

Example output (sample mode, abridged)

{ "type": "issue", "issue": "broken-link", "severity": "high", "sourcePage": "https://demo-links.siteverdict.test/",
  "target": "https://demo-links.siteverdict.test/missing-page", "status": 404, "anchorText": "Missing page",
  "fix": "Update the link to a working URL, remove it, or restore/redirect the missing page." }
{ "type": "issue", "issue": "missing-anchor", "target": "https://demo-links.siteverdict.test/guide#does-not-exist", "detail": "#does-not-exist not found on target" }
{ "type": "issue", "issue": "unverifiable-blocked", "severity": "info", "target": "https://partner.siteverdict.test/members", "status": 403 }

What it checks

Check	Notes
Broken links (4xx/5xx)	Internal and external <a>/<area> links. A HEAD failure is confirmed with GET before it is reported.
Dead domains, refused connections, TLS errors	Reported as broken, with the reason.
Broken resources	<img> + srcset, scripts, stylesheets, icons, iframes, video/audio/source, <object>, posters.
Redirect chains (2+ hops) and loops	Every hop is listed.
Internal links to redirecting URLs	So you can link to the final URL.
Missing #anchors	Same-page and cross-page fragments, checked against ids and <a name> on the target. SPA routes (#!/, #/) and text fragments are skipped. External fragments are optional.
Mixed content	Active (script, CSS, iframe: blocked by browsers) vs passive (images/media).
http:// links to your own https site	Insecure internal links.
Unverifiable (401/403/429/999)	Listed separately and never counted as broken. Many sites block automated checks, so check these manually.

Input highlights

startUrls, crawl (on by default), maxPagesPerSite (300, up to 10,000), checkExternalLinks, checkResources, checkFragments, checkExternalFragments, maxLinksPerSite (5,000), include/exclude patterns, requestDelayMs (500 ms for your site), externalDelayMs (1 s per external host), monitoring, alertWebhookUrl (secret), sampleMode.

Output

• Dataset views: Problems by source page (type: issue), Pages with problems (type: page), Link targets (type: link, non-OK only unless outputAllLinks), Summary.
• Key-value store: REPORT (HTML grouped by source page), REPORT-PDF, ISSUES-CSV, SUMMARY.

Pricing (pay per event)

Event	Price	When
page-scanned	$0.0005	Each HTML page fetched and scanned for links
link-checked	$0.0003	Each unique link target checked. Targets skipped by robots.txt, targets refused as private/local/internal addresses (also on a redirect hop) and links on crawled pages are free; each link row shows charged.
site-report	$0.01	Once per site: HTML/PDF/CSV report, monitoring diff and alert

Examples: a 200-page site with 1,500 unique links ≈ $0.56. A 2,000-page site with 10,000 links ≈ $4.01. Nothing is charged in sample mode or for errors on our side. The run stops cleanly at your max-charge limit: the status message then starts with "Partial (max total charge reached)", each start URL that was not reached gets a free error row, and a site whose report did not fit in the budget has reports: null with a reportsNote (the report is not charged).

FAQ

Will it hammer other people's sites? No. Each external host gets its own minimum spacing (1 s by default). Its robots.txt is fetched and obeyed, and retries back off. Links on hosts that disallow bots are reported as robots-skipped and are never requested.

Why is a LinkedIn/Amazon link "unverifiable"? Those sites answer automated requests with 999/403. We never try to get around that, so we mark them for manual checking instead of falsely calling them broken.

JavaScript-rendered links? Not followed. The checker reads the HTML your server sends, like most crawlers.

Limits

• 10,000 pages and 50,000 unique link targets per site, 20 sites per run.
• External #fragment checks download the target page (up to 2 MB).
• Soft-404s on external sites (a 200 "not found" page) cannot be detected reliably and are not flagged.

Sources & compliance

• Data: the public pages of the site you choose, and HTTP status responses of the URLs it links to. Fetched with the honest user agent SiteVerdictBot/1.0 (+hello@webrevenuescan.com). robots.txt (RFC 9309) is obeyed on every host and every redirect hop. Checked 11 Oct 2026.
• Never bypasses CAPTCHAs, logins, paywalls or bot protection. No proxies.
• SSRF-safe: private, loopback and metadata addresses are refused, and every redirect hop is re-checked.
• No personal data is collected. Anchor text is stored only for links with problems. If your pages contain personal data, you are the controller of the audit (UK GDPR/PECR). Only check sites you own or may audit.
• The webhook URL is a secret input, redacted from logs and never output.