blog-cannibalization

Run the analysis on your blog directory to extract primary keywords from titles, headings, and meta descriptions of each post. Posts are clustered by exact, stem, semantic, and subset keyword matches, and each cluster with two or more posts is flagged with a severity level and recommendation.

Use the default local mode, which analyzes file content directly via grep and read operations with no API keys required. It scores severity using heuristics like exact primary keyword matches and shared H2 keywords instead of SERP data.

Yes, API mode requires DATAFORSEO_LOGIN and DATAFORSEO_PASSWORD set as environment variables, read by a local CLI wrapper. If credentials are missing or the wrapper is unavailable, the analysis automatically falls back to local mode.

The analysis scans Markdown (.md), MDX (.mdx), and HTML (.html) files in the target directory. Files inside node_modules, .git, and drafts folders are skipped automatically.

The recommended action depends on severity: merge thin overlapping posts with a 301 redirect, differentiate posts serving distinct intents by retargeting one to a long-tail keyword, or add a canonical tag when one post is clearly the authority.

Analysis requires at least two blog posts in supported formats; a single post or empty directory exits with an explanatory message. Check that the path is correct and files use .md, .mdx, or .html extensions.