Codearia Academy

geo-seo-claude

Checks whether a site is ready to appear in answers from ChatGPT, Perplexity and AI Overviews. Seventeen terminal commands, a report in a file, MIT licence.

10kcommit this monthchecked August 31, 2026
geo-seo-claude cover: a robotic hand and a site audit running in the terminal

geo-seo-claude is a Claude Code skill that looks at your site the way a language model does. Ordinary SEO tools count positions in the results page; the question here is different — will ChatGPT cite you when someone asks about your subject.

The main command, `/geo audit`, splits the work across five subagents: AI visibility, platform-specific optimisation, technical foundations, content quality and E-E-A-T, structured data. A few minutes later you get a markdown report with a score from 0 to 100 and a prioritised task list.

In day-to-day work the short commands earn their place faster. `/geo crawlers` takes seconds to show whether robots.txt lets GPTBot, ClaudeBot and PerplexityBot through — the most common breakage, and one that makes any work on the text pointless. `/geo llmstxt` builds an llms.txt file from your sitemap. `/geo citability` cuts a page into blocks and scores each one: is there a direct answer up front, does the paragraph stand on its own, are there numbers and names. It works as an editorial exercise even if you never look at the score.

Worth taking on its own: `schema/` holds six ready JSON-LD templates — organisation, local business, product, article with an author, SaaS, site search.

One caveat about languages: the citability logic is built on English patterns, so the number on a Russian or Hebrew page is scored blind. Read the score on the English version and use the written recommendations elsewhere. The PDF report only builds on macOS; the markdown report works everywhere.

Open on GitHub

Comments