seo-sitemap has two modes. Analysis runs sitemap_discovery.py to read the Sitemap lines in robots.txt, probe common paths, follow an index one level down and count every entry, then reports limits, lastmod quality, cross-host and duplicate URLs and redacted tokens.
Generation asks for the business type, loads an industry template from the plan skill, applies location page quality gates and writes valid XML, split into an index at 50,000 URLs or 50MB, with a STRUCTURE.md. Image, video and news sitemap rules are covered.
What You Get
- VALIDATION-REPORT.md with an issues list and recommendations
- sitemap.xml or split files with a sitemap index
- STRUCTURE.md describing the URL organization
- Per-file counts for URLs, lastmod coverage and duplicates
Common Use Cases
- Find out why Search Console reports a sitemap problem
- Split a 100,000 URL sitemap into an index
- Remove noindexed or redirected URLs from a sitemap
- Build a first sitemap for a new site
What Makes It Unique
Counting and parsing are done by a helper script, so the agent does not tally URLs by hand. The helper uses a fetch layer that guards against requests to unsafe hosts, handles gzip sitemaps, and redacts tokens in listed URLs. Lastmod is judged for accuracy, since Google only trusts it when it is reliable.
Under The Hood
The skill runs the plugin’s sitemap_discovery.py through a launcher, which fetches robots.txt and sitemap files over the network and prints JSON that the agent reads. The agent also spot-checks a sample of listed URLs for HTTP 200 status. In generation mode it writes sitemap XML and a documentation file to disk and reads templates from the plan skill’s assets folder. The frontmatter declares none of this access. Fetched content is to be treated as untrusted data.