Claude SEO Skills: 39 free skills that turn Claude into a working SEO team

Built by the team behind 500+ client campaigns at StudioHawk. Install in about a minute. MIT licensed, no email gate, no paid tier.

Claude SEO Skills is a free, MIT-licensed library of 39 skill files that turn Claude into an SEO checker, auditor and strategist. Created by Lawrence Hitches and the StudioHawk team, each skill installs into Claude Desktop or Claude Code in about a minute. The whole library lives on GitHub.

Created by Lawrence Hitches ยท github.com/lhitches

claude-code-seo
$ claude "run the claude code seo audit"

Scanning HTML files in this repo...
589 files found, 15 page sample audited

Score: 90/100
A. Structural baseline    35/40
B. AI search readiness    18/20
C. Linking and structure  19/20
D. Content basics         14/15
E. Infrastructure          4/5

1 High  3 Medium  6 Low findings
Report written to SEO_REPORT.md

A skill is a process. A prompt is a guess.

A prompt asks Claude to improvise. A skill hands Claude a process: the exact steps, checks and output format a senior SEO would follow, written down and installed.

Most "Claude SEO" advice is a folder of prompts. Prompts drift. Ask the same question twice and you get two different audits, neither of which you can hand to a client.

Every skill in this library encodes how StudioHawk actually runs the task across 500+ client campaigns. Same input, same process, same report format, every time.

A Claude skill is a markdown file that teaches Claude a repeatable workflow: when to trigger, which steps to follow, what to check and what to output. A prompt is a one-off instruction. A skill runs the same process every run, so your tenth audit is as rigorous as your first.

What the skills do

Six categories. Every skill is a working process, and each card below tells you what it finds.

Audit and diagnose

14 skills

The SEO checkers. Technical audits, website scores, trust checks, schema generation and cannibalisation detection. The analysis tools that find what is broken and rank what to fix first.

Industry-specific audits

6 skills

Ecommerce, B2B, SaaS, local and restaurant audits. Plus an agentic product page auditor that checks whether an AI shopping agent can actually buy from your page.

Content and topical strategy

8 skills

Content briefs, topical authority maps, keyword clustering and information gain analysis. The planning layer that runs before anyone writes a word.

AI search

5 skills

AI visibility checks, citation monitoring, Clarity bot auditing and Claude Code SEO. The skills aimed at getting you cited by ChatGPT, Claude and Perplexity.

Outreach, links and reviews

2 skills

Data PR pitches and Google review responses, written the way an agency writes them.

Calculators and helpers

4 skills

Ad spend comparisons, calculator page specs, GBP reporting and a quick diagnostic for whether a Google update helped or hurt you.

Install a skill in about a minute

Three ways in. Clone the whole library, pull one skill with curl, or paste a skill straight into a chat.

1. The whole library (Claude Desktop, macOS/Linux)

git clone https://github.com/lhitches/claude-seo-skills.git
mkdir -p ~/.claude/skills/
cp claude-seo-skills/skills/*.md ~/.claude/skills/
cp -r claude-seo-skills/skills/topical-authority-map ~/.claude/skills/

2. One skill with curl (example: Claude Code SEO)

curl -o ~/.claude/skills/claude-code-seo.md https://raw.githubusercontent.com/lhitches/claude-seo-skills/main/skills/claude-code-seo.md

3. No install at all

Open any skill file on GitHub, copy the contents, paste it into Claude as your first message. The process runs the same way.

Skills run on every Claude plan, Claude Code included. Windows commands are in the repo README.

Real output, real site, no mockup

This is genuine output from the Claude Code SEO skill run against lawrencehitches.com, my own site. No mockup. The run was not rigged to look good.

SEO Audit Report: www.lawrencehitches.com

Score: 90/100. 15 of 589 HTML files audited as a representative sample. All 15 pages have correct self-referencing canonicals, full Open Graph sets, valid JSON-LD and zero broken internal links.

  • One real content bug: the Officeworks case study ships the wrong meta description. Line 7 reads "Why can't Google crawl my React website?", which describes a different article entirely, and the same wrong text repeats in og:description and twitter:description.
  • Recurring drift: 5 of 15 titles exceed 60 characters (longest is 105) and 4 meta descriptions fall outside the 120 to 160 character window.
  • Inconsistent state: 5 category pages are indexable but appear in none of the 4 sitemap files. Pick one: in the sitemap and indexable, or noindexed and excluded.
#SeverityFileIssue
1Highcase-study-officeworks/index.htmlMeta, og and twitter descriptions all describe an unrelated React article
2Mediumsitemap-core.xml5 indexable category pages absent from every sitemap file
3Mediumai-seo-tools/index.htmlTitle is 105 characters, well past the 60 character display limit
4Lowanswer-engine-optimization/index.htmlInternal link resolves through a 301 hop; point it at the destination

Excerpt from SEO_REPORT.md, generated 2026-07-09. Full format: score header, top 10 fix-forward table, findings by category, page rankings.

It found real issues on my own site, and that is the point. The skill does not flatter the person who wrote it, and it will not flatter your site either.

SEO, GEO and AEO: what each skill layer targets

Modern search is three layers. Google rankings, AI citations and direct answers. Most tasks touch all three, and the table maps each layer to the skills that cover it.

LayerGoalOptimises forKey signalsSkills that cover it
SEO Rank in Google and Bing Crawlers and ranking systems Technical health, internal links, E-E-A-T, intent match Technical SEO Audit, Website Score, Striking Distance Finder
GEO Get cited in AI answers Retrieval and grounding systems Entity clarity, structured facts, provenance, consistency AI Visibility, Entity SEO Auditor, AI Citation Monitor
AEO Win the direct answer Question and answer extraction Answer-first passages, schema, self-contained chunks Decision Content Builder, Schema Markup Generator, Citation Share Auditor

All 39 skills

Filter by category. Every card links to the skill file on GitHub, and flagship skills have a full page with install commands and sample output.

Audit and diagnose

Author Authority Builder

Turns authors from byline strings into entities: author page, Person schema with sameAs, byline fixes, cross-web consistency, three bio lengths. The E-E-A-T person layer.

Skill file on GitHub
Audit and diagnose

Cannibalization Detector

Finds the queries where two of your own pages compete and split the signal. Picks the keeper and gives one fix: consolidate, canonical, or differentiate.

Skill file on GitHub
Audit and diagnose

Content Decay Detector

Finds pages quietly losing traffic, diagnoses why (freshness, displacement, intent shift, demand), and prescribes refresh, rewrite, consolidate, or leave.

Skill file on GitHub
Audit and diagnose

Entity SEO Auditor

Audits whether Google's Knowledge Graph and AI search recognise your brand as an entity. Entity home, schema, sameAs, Wikidata presence, cross-web consistency. One fix per layer.

Skill file on GitHub
Audit and diagnose

Helpful Content Audit

Grades any page against Google's 32 helpful-content self-assessment questions: evidence per verdict, YMYL-aware weighting, letter grade, prioritised fixes.

Skill file on GitHub
Audit and diagnose

Search Intent Mapper

Maps keywords to intent, builds one-cluster-one-page groups, names the content type each needs, and flags pages fighting their query's intent.

Skill file on GitHub
Audit and diagnose

Technical SEO Audit

Finds the technical issues stopping Google from reading your site. Redirect chains, duplicate titles, indexing problems, orphan pages, cannibalisation.

Skill file on GitHub
Industry-specific audits

Agentic Product Page Auditor

Audits whether an AI shopping agent can read and buy from your product page. Scores the product graph across six layers and gives one template-level fix per problem, ranked by what blocks a sale.

Skill file on GitHub

No skill matches that search. Try a broader term, or browse the repo.

Questions SEOs actually ask

Is Claude actually good at SEO in 2026?

Yes, for process work. Audits, briefs, keyword clustering, schema, reporting. Claude is only as rigorous as its instructions, which is exactly why this library exists. A skill gives Claude a senior SEO's checklist, so the output holds up in front of a client instead of reading like chatbot filler.

What is the difference between a Claude skill and a prompt?

A prompt is a one-off instruction that Claude interprets fresh each time. A skill is an installed markdown file with trigger conditions, steps, checks and a fixed output format. Prompts drift between runs. Skills produce the same report structure every run, which is what makes them usable on client work.

Should I use Claude or ChatGPT for SEO?

Both handle SEO tasks. This library targets Claude because skills give it an install-once workflow layer, and Claude Code can audit an entire website codebase from the CLI. If your team runs ChatGPT, the skill files still work as structured instructions: paste the contents in and the process holds.

Which SEO tasks can I hand to Claude, and which still need a human?

Delegate the repeatable work. Crawl analysis, technical audits, content briefs, clustering, schema generation and title rewrites. Keep a human on strategy, prioritisation trade-offs and anything that ships to production. Every audit skill here outputs findings for review. None of them auto-publishes changes to your site.

Do these skills replace an SEO specialist or an agency?

No. They compress the hours a specialist spends on process work. A senior SEO runs faster with them, and a founder gets agency-grade structure without the retainer. StudioHawk still puts humans on every one of its 500+ campaigns. The skills carry the checklist, people carry the judgement.

Does Claude keep up with Google algorithm updates?

No model reads yesterday's update. Claude's training has a cutoff date, so the skills are built around durable process instead of tactics that expire: crawl logic, entity audits, intent mapping. When an update hits, the Winner or Loser skill diagnoses the impact from your own Search Console data.

Do I need a paid Claude plan to use these?

No. Anthropic makes skills available on every Claude plan, Free included, as long as code execution is enabled in your settings. Skills also run in Claude Code. If you would rather not install anything, paste a skill file straight into a chat and the same process runs manually.

Created by Lawrence Hitches

I am Chief of Staff at StudioHawk, Australia's largest dedicated SEO agency, where I have worked across 2,000+ SEO campaigns with a 120-person team. These skills are our internal processes, open-sourced. More than 6,000 students have taken the free Hawk Academy course they grew out of.

The library is MIT licensed and stays free. No email gate, no premium tier, no upsell.

Get the skills on GitHub