SEO Technical Auditor
Audit a website's source or built output (Next.js, Astro, or static HTML) for technical on-page and crawl hygiene: title and meta description length and uniqueness, self-referencing canonicals, one H1 and an unbroken heading hierarchy, presence of JSON-LD, sitemap and robots.txt, redirect chains, internal-link click depth, and image alt text, then return one findings table ordered by severity with the file or route to fix. Use when you want the crawl and markup plumbing checked across a whole site, not the quality of the writing or the correctness of the structured data.
Point this skill at a Next.js, Astro, or static-HTML project or its build folder and it extracts every page's title, meta description, canonical, robots directive, headings, JSON-LD, images, and internal links, checks them against fixed rules, diffs the sitemap against real routes, traces redirects, and returns a findings table by severity. Read-only.
Install to ~/.claude/skills/seo-technical-auditor/SKILL.md
npx agentscamp add skills/seo-technical-auditorTechnical SEO problems are boring and expensive: a canonical that still points at the staging host, a noindex left over from a preview deploy, a sitemap full of redirected URLs, three hundred pages sharing one meta description. This skill audits the plumbing. Point it at a repository (Next.js, Astro, or plain HTML) or at a built output folder and it extracts every page's title, meta description, canonical, robots directives, headings, JSON-LD blocks, images, and internal links, checks each against the rules below, and returns one findings table ordered by severity with the file or route to fix. It reads files and runs read-only commands; it does not edit pages.
When to use this skill
- Before a launch or a domain, framework, or URL-structure migration, and again right after.
- Search Console reports "Duplicate without user-selected canonical", "Crawled, currently not indexed", or "Excluded by noindex" on pages that should rank.
- A monthly hygiene pass on a content site that adds pages faster than anyone reviews them.
- You inherited a site and need a baseline before touching any content.
When NOT to use this skill
- You want to know whether a page is written so an AI answer engine can quote it: the geo-content-auditor skill. This skill checks that JSON-LD exists; it does not judge the prose.
- You need JSON-LD written or validated property by property: the schema-markup-generator skill.
- The complaint is speed or layout shift: the web-vitals-optimizer skill works from field data.
- One page, two minutes: the /seo-check command runs a subset of these checks on a single file or URL.
- You want an llms.txt file: the llms-txt-generator skill.
Instructions
-
Choose the surface and say which one you audited. Built HTML is the truth; source is an approximation, because frameworks compute metadata at build time. If
npm run build(orastro build) is cheap, run it and auditout/,dist/, or.next/server/app/**/*.html. Otherwise read the source:metadataandgenerateMetadataexports plusapp/sitemap.tsin Next.js, the<head>of Astro layouts, the raw<head>of static HTML. Record the surface at the top of the report. -
Inventory routes from two directions. Parse
sitemap.xml(or the sitemap index) for the declared set, and discover the actual set withfind out -name '*.html'or the route tree. Diff them: pages missing from the sitemap, sitemap entries that 404 or redirect, and sitemap entries that carrynoindex. -
Extract per-page fields. For each page pull
<title>,meta[name=description],link[rel=canonical],meta[name=robots], everyh1toh6in order, eachscript[type="application/ld+json"], everyimgwith itsalt, and every internal<a href>. A short Node script with an HTML parser is more reliable than regex, butrgis enough for a first pass:rg -o --no-filename '<title>[^<]*</title>' out | sort | uniq -c | sort -rn | head # duplicate titles rg -L -l 'rel="canonical"' out --glob '*.html' # pages with no canonical rg -o '<img[^>]*>' out | rg -v 'alt=' # images with no alt attribute -
Apply the page rules. Title 50 to 60 characters and unique across the site; meta description 120 to 155 characters and unique; canonical present, absolute, on the production host, and self-referencing unless the page is a deliberate duplicate; exactly one
h1; no heading level skipped; nonoindexon a page that is in the sitemap; every JSON-LD block parses as JSON. Count each rule's failures and keep three example routes per rule. -
Check robots.txt and headers. Confirm
robots.txtexists, lists aSitemap:line, and does notDisallow: /or block assets the pages need to render. If you were given a live URL,curl -sI https://example.com/some-page | rg -i 'x-robots-tag|location|content-type'catches header-levelnoindexand redirects that the HTML cannot show. -
Trace redirects. Read the redirect config (
next.config.*,vercel.json,_redirects,.htaccess), then sample a dozen old URLs withcurl -sIL -o /dev/null -w '%{http_code} %{url_effective}\n' <url>. Flag chains longer than one hop, temporary (302/307) redirects for permanent moves, and inconsistent trailing-slash orwwwhandling. -
Measure internal-link depth. Breadth-first search from
/over internal links. Report pages more than three clicks from the home page, and orphans: pages present in the sitemap with zero inbound internal links. -
Write the findings table. Severity is a rule, not a feeling. High: anything that stops indexing or sends it to the wrong URL (noindex on sitemap pages, canonical to another host, robots blocking, redirect loops, 404s in the sitemap). Medium: duplicate or missing titles and descriptions, missing
h1, redirect chains, orphan pages. Low: lengths outside range, missing alt text, skipped heading levels.
NOTE
Presence is all this skill checks for structured data: that a JSON-LD block exists and parses. Whether its properties are right, and whether the marked-up content is visible on the page, is the job of schema-markup-generator.
WARNING
Do not mass-edit titles and descriptions from this report. Each one is a content decision that changes what a page promises in search; the report names the pages, and a person or the seo-specialist agent decides the wording.
Output
A Markdown report with: the surface audited and the page count; a table with columns Severity, Rule, Failures, Example routes, and Where to fix (a file path or component); the sitemap diff; the redirect sample; and the orphan list. It ends with the three fixes that remove the most High findings.
Filed under
claude · seo · technical-seo · marketers · developers
Related
- Schema Markup GeneratorWrite JSON-LD structured data (Article or TechArticle, FAQPage, HowTo, SoftwareApplication, Product, BreadcrumbList, Organization) from the content that is actually visible on a page, fill the required and recommended properties from Google's structured-data documentation, link the nodes in one @graph, and validate the result with the Schema.org validator and the Rich Results Test before it ships. Use when a page needs structured data written, or an existing block checked property by property.
- SEO SpecialistUse this agent to review a site or a set of pages for search and answer-engine performance and get one prioritized plan: it checks technical hygiene with the seo-technical-auditor skill, the match between each page's title, H1, and lead and the query it targets, title and meta wording, internal linking, answer-first structure with the geo-content-auditor skill, and structured data with the schema-markup-generator skill, then ranks the fixes by effect and effort. Read-only; it never edits a page or invents a search volume. Examples: "review our docs site for SEO and tell me what to fix first", "why does this guide rank but never get cited by AI Overviews", "audit the ten pages in this folder and give me a plan".
- GEO Content AuditorAudit one published page or draft for answer-engine and LLM-citation readiness: an answer-first lead, question-shaped headings each answered in their first sentence, one fact per sentence, figures with a date and a source beside them, definitions that can be quoted whole, consistent entity names, and a visible FAQ block, then return a pass-or-fix scorecard and an ordered rewrite plan with replacement text. Use when you want to know why a page is not being quoted by AI Overviews or chat assistants and what to change in the prose, not in the markup or the crawl setup.
- llms.txt GeneratorGenerate or refresh a site's llms.txt (and, on request, llms-full.txt) from its sitemap, route tree, docs navigation, or content directory, following the llmstxt.org conventions: an H1 with the site name, a blockquote summary, H2 sections of links each with a one-line description taken from the page's own metadata, and an Optional section for secondary pages. Use when a site or docs repository needs the file created, or an existing one checked against the pages that actually exist today.
- 9 Best Claude Skills for SEO (and AI Search)Compare Claude skills, commands, and agents for technical SEO audits, structured data, answer-engine readiness, llms.txt, content briefs, landing pages, Core Web Vitals, and brand voice.
- SEO CheckRun a single-page SEO and AEO check on a file or URL (title and meta lengths, one H1, heading order, answer-first lead, FAQ, structured data, canonical, internal links, alt text) and return a pass-or-fix table.
- An SEO Content Workflow in Claude Code: Brief, Draft, On-Page CheckA repeatable SEO content workflow in Claude Code: a content repo with CLAUDE.md, SERP research over Exa and Firecrawl MCP, a brief, a draft, and on-page checks.
- llms.txtGenerate or refresh llms.txt from the project's routes, sitemap, or docs navigation, write it to the static directory, and verify that every link resolves.