blog-cannibalization

This skill detects keyword cannibalization across a set of blog posts, that is, when multiple pages compete for the same search intent and dilute each other's rankings. It solves an SEO content-strategy problem by extracting the primary and secondary keywords from each post's title, headings, first paragraph, and meta description, clustering semantically similar targets, and flagging pages that overlap. The output is a severity-scored report with concrete recommendations such as merge, differentiate, or canonicalize.

It operates in two modes. The default local mode is free and read-only: it uses Glob, Grep, and Read to scan Markdown, MDX, and HTML files, tokenizes and normalizes keyword phrases, and clusters posts using exact, stem, semantic-intent, and subset matching rules, then applies a local severity heuristic. An optional API mode (enabled with a flag) enriches the analysis with DataForSEO's Page Intersection and Ranked Keywords endpoints for SERP-level data at roughly one cent per call, computing a severity score from overlap count, search volume, and ranking-position gap. API mode requires DataForSEO credentials supplied through environment variables and a dedicated local CLI wrapper; the skill is explicit that credentials must never be exposed in prompts or reports and that it should fall back to local mode if no wrapper is present.

The skill targets content marketers, SEO specialists, and technical bloggers managing a corpus of posts. Typical use cases include auditing a blog for competing pages, deciding which thin posts to merge, prioritizing canonical tags, and quantifying the urgency of overlaps before a content consolidation project.