Site X-Ray: a website audit that ranks what to fix and shows the proof

Ask it in your own words

• "Audit https://example.com, 300 pages max."
• "We migrated last month. Can you check nothing broke?"
• "whats wrong with my site, it doesnt rank"
• "How is the leading site in my niche set up?"

What happens next

1. You give a URL. That is enough to start. The rest is asked in one message while the crawl runs.
2. It crawls politely. crawl_audit.py, the crawler, reads your pages at about two requests a second, follows robots.txt and stops at a page cap.
3. It checks before it reports. Claude opens examples of each finding, merges those that share one cause and drops false positives with a note.

What you get back

• An executive summary in five lines
• A score per area, with the share of the rubric that could be measured
• The fix-first list: the evidence, the exact fix, a done-when check
• What is already good, and what could not be measured
• A 30-day action plan, findings.csv and a fix tracker

Run on this very site

A full crawl of the five pages the site had that day.

Score 92/100: Healthy technical basics on the pages examined. Coverage 80% of the rubric.
CRAWL-HOSTS | high | https://www.seoskills.click/ -> first redirect is HTTP 307 (temporary); use 301 or 308
CRAWL-SITEMAP | low | No usable sitemap. robots.txt declares 0; tried: /sitemap.xml -> 404 [...]
OG-BASICS | low | 5 of 5 page(s) lack some of og:title, og:description, og:image, twitter:card; e.g. missing og:image

All three were fixed the same day: permanent redirects, a sitemap and a robots.txt, sharing images. The next run read 98.3/100.

What it will not do

• It sees the site, not Google: indexation, queries and rankings need Search Console. A clean score can sit next to a site that is mostly not indexed.
• It runs no JavaScript: text a script writes into the page has to be checked in a browser.
• If the site refuses the crawler (401, 403 or 429), it stops and says so. It never works around a block.
• The score measures technical hygiene, not future rankings, and never reads 100 when a check failed.

What it needs

Claude Code, Claude.ai, Claude Desktop or Cowork. The crawler needs code execution with Python 3.9+ and outbound web access. Without them, a manual route audits 10 to 15 pages chosen by template.

How to install a skill in Claude

Guides

• How to use Claude for SEO: which Claude for which job, and what data to give it
• Claude skills for SEO in 2026: seven options, and which one fits your situation
• How to install a skill in Claude, and what to check when it does not load
• Are Claude skills from independent developers safe to install?
• Discovered - currently not indexed: what it means and what to do, in order

Where it hands over

• Index Doctor: When the score is high and pages are still missing from Google, the cause lies outside what a crawl sees.
• Link Weaver: The audit counts orphans and weakly linked pages; pillar and cluster links are designed there.
• True North: After reading how a leading site in your niche is built, it turns what you saw into a plan sized to your hours.

Questions about this skill