claude-seo-geo

Claude SEO GEO

SEO and GEO skills that work like a specification: they audit a page, compare it with the competitors that rank, and build the fix. Rank in Google and get cited by ChatGPT, Perplexity, Gemini and Google AI Overviews.

27 skills, 13 zero-dependency scripts, every threshold either measured (with its source) or labeled as a field heuristic from 115+ real agency audits. Free and open source.

How a page skill works

Every skill that owns a page type (homepage and sales page, service, product, collection, comparison, blog article, local page, and the 21 types of the page sections skill) runs the same four phases, in the same order, with the same deliverable.

The finish line is measurable: no requirement left failing, or a written owner decision for each one that stays.

What the benchmark looks like

$ python3 skills/seo-geo-audit/scripts/page_benchmark.py --type homepage \
    https://your-site.com/ https://rival-a.com/ https://rival-b.com/ https://rival-c.com/

| metric      | your-site.com | rival-a.com | rival-b.com | rival-c.com |
| words       | 847           | 1325        | 2187        | 2499        |
| seo_score   | 80            | 84          | 54          | 88          |
| bold        | 0             | 1           | 34          | 6           |
| money_links | 1             | 15          | 10          | 15          |
| schema      | -             | Organization, Person, WebSite | - | EducationalOrganization, WebSite |
...
BLOCKS half or more of the competitors have and the client does not:
  - author_block (2/3)  - guarantee (2/3)  - contact_details (2/3)
CLIENT BELOW THE COMPETITOR MEDIAN:
  - words: client 847 vs median 2187
  - bold: client 0 vs median 6

Site level: crawl and decay

python3 skills/seo-internal-linking/scripts/site_crawl.py https://your-site.com --out crawl.json
python3 skills/seo-traffic-drop/scripts/content_decay.py --current last3m.csv --previous prev3m.csv --last-year ly3m.csv --crawl crawl.json

The crawl finds what no single-page audit can see (orphans, internal links pointing to redirects, cannibalization, pages buried deeper than 3 clicks). The decay finder turns three Search Console exports into a refresh queue: which pages lose traffic, why, and what to do.

What the audit measures

seo_audit.py uses the same thresholds and weights as the free Sorank Chrome extension, so the terminal and the browser agree on the same page. Per page it returns an SEO score and a GEO score out of 100 and a ranked list of findings, covering:

• Tags and head: title and meta lengths, H1 and heading tree, canonical, Open Graph, Twitter card, viewport, lang, charset, favicon, hreflang, meta refresh, robots and snippet directives.
• Content: word count, bold usage, keyword density (alert above 2.5%, stuffing above 7%), the keyword placement matrix (title, H1, H2-H6, meta, URL, intro, bold, alt), readability with sentence-length buckets and Flesch, paragraph length, question headings, statistics, definitions, quotes.
• Links: internal and external, nofollow, sponsored and ugc, empty and generic anchors, body links to the pages that sell, social profiles and placeholder links.
• Images: alt coverage, alt quality, weight, format, dimensions, lazy-loading of the first image.
• Structured data: types found, JSON-LD that fails to parse, required properties missing or empty per type.
• Templates: table of contents, author block and author page link, dates, breadcrumb, CTA links, blog hub cover images.
• Trust: income or results promises without a disclaimer, gambling topics without the legal notice, unbacked superlatives.
• Site level: robots.txt rules for AI search, user-fetch and training bots, sitemap and its 50,000 URL limit, llms.txt.

The skills

Page skills (spec, audit, benchmark, build)

Site skills

Authority, AI visibility and distribution

YouTube videos (any sector)

The scripts

All standard-library Python 3.9+, read only, no API key, no install.