broken-link-checker is a technical SEO workflow for AI agents. Given a site URL, it inspects the links on that page and a bounded sample of related pages, or it analyzes a crawler export the user supplies.
Findings are grouped into confirmed internal 404s, external 404s, server errors, redirect issues and inconclusive checks. Each broken target is listed with its source page and a specific fix. The report says how many pages were inspected, so a clean sample is never presented as a clean site.
What You Get
- A list of confirmed internal and external 404s with the page each one appears on
- Redirect issues with their destinations and check times
- Inconclusive checks kept apart from confirmed errors
- A specific fix per target: update the link, restore the page or add a redirect
- A statement of pages inspected and coverage limits
Common Use Cases
- Checking a site for 404s before a relaunch
- Diagnosing a drop in crawl health from dead internal navigation
- Reviewing a crawler export and turning it into a fix list
- Running a quick website health review for a client
What Makes It Unique
No crawler is bundled, so the skill reports its coverage openly: it samples up to 50 pages by default, files timeouts, blocked fetches, login pages and rate limits under inconclusive checks, and never invents HTTP statuses when a tool only exposes page text. It also refuses to follow private, loopback, link-local or credential-bearing URLs collected from a page.
Under The Hood
The skill is a short wrapper whose SKILL.md tells the agent to read a longer WORKFLOW.md in the seo folder of the same repository and follow it. The workflow is prose only and ships no scripts. It has the agent fetch public web pages with whatever browser or fetch capability the host offers, and follow only public HTTP and HTTPS links. It records status, redirect target and check time for each link. It does not edit links or publish redirects until the user approves the exact change, and it treats fetched page text as untrusted data.