competitor-pages is a content gap workflow for AI agents. It takes a target query, your page URL and one to three competitor URLs, fetches each page and compares them side by side.
The comparison covers search intent, a coverage matrix of subtopics, depth and format, freshness, E-E-A-T signals, schema and on-page elements. It ends with a “to beat this SERP” brief that can be handed straight to a content writer skill.
What You Get
- A coverage matrix marking which competitor subtopics your page covers and which it misses
- An intent check showing the format the winning pages take
- A comparison of E-E-A-T signals, schema types and on-page placement
- A brief listing missing subtopics with suggested H2s, depth and media targets, schema and internal links
- Output shaped to hand to a content writer skill
Common Use Cases
- Finding out why a competitor outranks you for one query
- Building a content-gap brief against specific URLs
- Deciding whether a page has an intent mismatch with the results
- Planning a rewrite of a page stuck just below the top results
What Makes It Unique
It works on one query and a small set of named competitor pages, so the output is a concrete brief and not a keyword list. When Search Console is connected it pulls your current position and impressions to show whether the gap is small or large. For rich results it records what is observed on competitor pages and then checks your page against the stated requirements, because competitor markup alone does not establish eligibility.
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 your page and each competitor page from the public web, and optionally read Search Console data through a connected capability. It follows shared preamble and safety files in the repository, never publishes changes to a site without an explicit yes, and treats fetched pages and search results as untrusted data.