link-rot-detection

Broken external links in markdown are found by extracting URLs from `.md` and `.mdx` files and running HTTP health checks against them. This process classifies internal versus external targets to avoid false positives and flags unreachable references.

Link rot is the gradual degradation of external URLs becoming unreachable or dead over time. You detect it by parsing markdown content trees for external references and performing HTTP status checks to identify persistent failures like 404 or 410 errors.

Soft 404 pages are detected during a link audit by analyzing HTTP responses for "unknown page" interstitials even when the server returns a 200 status code. This prevents falsely validating links that point to placeholder error pages.

Yes, scheduled broken link checks are supported for periodic link-health maintenance workflows. The skill applies retry and backoff politeness for transient failures, making it suitable for nightly scans and CI-like reviews of content repositories.

A link health report for dead URLs is generated as both a JSON report and a markdown summary. These structured outputs record redirects, transient failures, and persistent dead links, making them suitable as audit artifacts.

Broken link detection handles HTTP redirects by recording them during health checks, and manages transient failures using retry and backoff strategies. This ensures links are accurately verified before being flagged as dead.