Broken Link Checker & Sitemap Validator (Monitor Mode)

Find broken links, 404s, redirects and sitemap errors. In monitor mode, get alerted only about new ones.

This broken link checker crawls your site, status-checks the links, images, scripts and stylesheets on the pages it crawls (up to maxLinksToCheck unique URLs), and validates your XML sitemap. A 4xx is confirmed with a GET request before it's reported, and 5xx, 408, 429 and network errors are retried. Bot-blocked links are re-checked in a real browser; if they still can't be verified, they're a warning, not an error.

Use cases

• SEO agencies: a weekly link and sitemap check per client, with an HTML report.
• Site owners: catch dead links after a migration or a deleted page.
• Developers: run it after a deploy and fail the pipeline when errors > 0.

Sample output (shortened, real run of the example input: web-scraping.dev with limits of 20 pages, 300 link checks and 50 sitemap URLs):

{
  "category": "link",
  "issueType": "broken-link",
  "severity": "error",
  "url": "https://web-scraping.dev/api/graphql",
  "statusCode": 404,
  "explanation": "HTTP 404 Not Found.",
  "foundOnCount": 18,
  "firstFoundOn": "https://web-scraping.dev/",
  "anchorText": "Graphql",
  "checkedWith": "HTTP GET"
}

That run took 27 seconds: 19 pages crawled, 108 link checks (85 page links + 23 sitemap URLs not already checked; the other 27 sitemap URLs reused earlier results at no charge). Cost: 19 × $0.002 + 108 × $0.0005 = $0.092.

No subscription. You pay $0.002 per page crawled and $0.0005 per link checked, only when it runs.

What does it check?

Result	What it means	How it's verified
broken-link	4xx such as 404, 410 or 400	HEAD first; any failure is re-checked with GET
server-error	5xx that keeps failing	Retried with backoff; Retry-After is respected
unreachable	DNS, connection or certificate failure	Re-checked; reason given in plain language
blocked-unverified	Possible bot protection (403, 429, 999, Cloudflare). Not proof the link is broken	Retried, then the first 25 (by default) are opened in headless Chrome. A browser 404/410 is reported as broken; the rest stay a warning
soft-404	Answers 200 but shows a "not found" page	Compared with your site's 404 response and "not found" titles in several languages
redirect, redirect-chain, redirect-loop	The link redirects, once, several times or in a loop	Every hop is recorded
login-required, malformed-url	HTTP 401 (info only); an href that isn't a valid URL	–

Each unique URL is checked once per run, including sitemap URLs already checked as links. Every row lists where the link was found (up to 25 pages) and how it was checked (checkedWith).

Sitemap validator

It finds sitemaps via robots.txt or common locations, follows indexes and .gz files, and reports:

• Sitemap URLs that are broken, redirect, are soft 404s, need a login or are blocked by robots.txt.
• noindex pages in the sitemap, which send search engines contradictory signals, and canonicalised URLs whose canonical tag points elsewhere.
• Orphan pages (in the sitemap, but nothing links to them) and pages missing from the sitemap.
• File problems: missing sitemap, no <urlset>/<sitemapindex> root, over 50,000 URLs or 50 MB uncompressed, bad <lastmod>, duplicates, a broken Sitemap: line in robots.txt.

It checks structure, not the full XML schema. Reading stops at maxSitemapUrls. URLs pointing to another host are flagged, not checked.

Input

Field	Description	Default
startUrls	Your site, usually the homepage. Other sites are checked, not crawled.	–
maxPagesToCrawl	Internal pages to crawl for links	100
maxLinksToCheck	Cap on unique URL checks (links, resources and sitemap URLs)	2,000
checkExternalLinks, checkResources	Also check links to other sites, and images, scripts and stylesheets	true
validateSitemap, sitemapUrls, maxSitemapUrls	Validate the sitemap; extra sitemap URLs; how many sitemap URLs to read	true, –, 500
verifyBlockedInBrowser, maxBrowserChecks	Re-check bot-blocked links in headless Chrome (free)	true, 25
renderJavaScript	Find links added by JavaScript (slower; use 2 GB+ memory)	false
reportRedirects	Which redirecting links to report: internal, all or none	internal
excludeUrlPatterns	Skip URLs matching these regular expressions or plain text	–
monitorMode, monitorStoreName, outputOnlyNewIssues, alertOn	Compare with the previous run; output or alert on new issues only	off, new-errors
webhookUrl	Slack, Discord, n8n, Make or Zapier webhook	–
notificationEmail	Optional email alert	–
maxConcurrency, requestTimeoutSecs, proxyConfiguration	Advanced	16, 20, none

The Console form is prefilled with smaller limits (20 pages, 300 link checks, 50 sitemap URLs) for a quick first try; the defaults apply when a field is empty. Several start URLs share one budget, so use one run or Task per site.

Output

One dataset row per issue, sorted by severity, then a category: "summary" row with the totals (pagesCrawled, linksChecked, errors, brokenLinks, crawlComplete, and in monitor mode newIssues and resolvedIssues). It's always added, even when nothing is wrong.

The key-value store holds report.html (a shareable report) and OUTPUT (the summary as JSON).

Monitor mode and scheduled checks

Monitor mode is off by default. For a scheduled check that reports only changes:

1. Turn on Monitor mode and save the input as a Task, one per site.
2. In Apify Console, open Schedules and add the Task with a cron expression, for example 0 6 * * 1 for Mondays at 06:00.
3. Add a webhookUrl. With alertOn: new-errors (the default), an alert is sent when a new error appears or when an earlier issue is resolved.

Each issue gets isNew and firstSeenAt. The first run is the baseline: every issue on it is new, so it alerts on every error. Set maxPagesToCrawl above your site's page count: when the limit is hit, different pages can be crawled each run and issues can flip between new and resolved. See Apify schedules.

Use it from n8n, Make and Zapier

Apify's official integrations can run a Task, get its dataset items, or start a workflow when a run finishes. For example:

• Client sheet: append the summary row's totals to a Google Sheet.
• Ticket per new broken link: with monitor mode on, filter rows with isNew: true and severity: "error" and create a Jira, Linear or Trello card. Skip the first (baseline) run, or it will open a card for every existing issue.
• After a deploy: run the Task and stop the release when the summary has errors > 0.

The webhook sends JSON with a text field (Slack, n8n, Make, Zapier) and a content field (Discord). Guides: n8n, Make, Zapier.

Pricing (pay per event)

Event	Price
Page crawled (page-crawled): an HTML page on your site downloaded and scanned for links	$0.002
Link checked (link-checked): each unique URL status-checked, including sitemap URLs	$0.0005

Crawled pages aren't also charged as link checks. Retries, browser re-checks, the summary row and the report are free.

Example: a 100-page site with 2,000 unique URLs (the 100 pages plus 1,900 other links, images, scripts and sitemap URLs) costs 100 × $0.002 + 1,900 × $0.0005 = $1.15 per check. That's about $5 a month weekly or $35 a month daily. With the default limits one site costs at most about $1.20 per run. Set Maximum cost per run for a hard cap; the Actor stops cleanly when it's reached.

FAQ

Will it report links that actually work? It tries not to, but bot protection makes some links impossible to verify. Those are warnings to check by hand; exclude sites that always block it.

I run a shop. Why are there so many URLs? Each query-string variant (?sort=, filters, ?add-to-cart=) is a separate URL and is charged separately. Add patterns such as \?sort=, \?filter or add-to-cart to excludeUrlPatterns.

Do I need email alerts? No, they're optional. If you enter an address, the alert is sent by Apify's apify/send-mail Actor on your own account, which uses a little platform credit.

Limitations

• Crawling stays on the start URL's site. Other subdomains are checked as links but not crawled.
• Links built by JavaScript are only found with renderJavaScript. Images referenced in CSS url() aren't checked.
• Every 403 is treated as possible bot blocking and reported as blocked-unverified, even when the page really is forbidden.
• Bursts of 403 or 429 answers from big external sites are reported as warnings. The Actor doesn't yet slow down automatically for those sites.
• Links are checked before sitemap URLs, so when maxLinksToCheck runs out, remaining sitemap URLs are skipped.
• Orphan pages are only reported when the whole site was crawled and the sitemap fully read; missing-from-sitemap needs a fully read sitemap.

Keywords

Broken link checker, dead link checker, 404 checker, sitemap validator, link monitor, redirect checker, soft 404 checker.