Every link checker produces a report and most reports go unread, because the output is undifferentiated and the obvious response is to fix all of it. Ranking by what actually matters — a broken link in the main navigation against one in a decade-old post — is what makes the report actionable. The monitoring half is the other half of the same argument: links break after you check, not while you are checking, so a one-off scan is a snapshot of a problem that recurs. The health check is a small honesty, keeping a backend failure from reading as a clean site.
A broken-link service in three parts: a one-off scan of a URL or an entire site, scheduled monitoring, and remediation suggestions derived from a report.
- check_broken_links scans a single URL or follows a sitemap across the whole site, returning a structured report.
- monitor_links sets up ongoing checking on a schedule rather than requiring you to remember.
- get_fix_suggestions takes a report and returns prioritised remediation rather than a flat list of failures.
- health_check confirms the server can actually reach its backend, so a clean scan and a broken connection are distinguishable.
Nothing named for the scan.
One command plus a key — pip install linkrescue-mcp, then supply credentials