pro-seo
The front door. Describe an SEO problem in plain English and it routes to the right specialist skill.
32 SEO skills for Claude Code — technical audits, content quality, schema, AI-search visibility, local, e-commerce, international, and a staged lifecycle that runs the whole programme end to end. This site is generated from the repository, so it documents exactly what is installed.
New here? The fastest useful thing is a single-page check — it needs no credentials and no setup beyond installing the plugin:
/pro-seo page https://example.com/
Then read Getting started for install and the three commands worth learning first.
Search by name, by what it does, or by the
problem you have — try schema, indexability or local.
The orchestrators. If you do not know which skill you need, run one of these and it routes for you.
The front door. Describe an SEO problem in plain English and it routes to the right specialist skill.
A staged programme rather than a one-off check. Use it when you want a plan that survives across sessions and remembers where it got to.
A full-site sweep. Spawns up to 15 specialists in parallel, so it is the slowest and most thorough thing here.
Point these at a URL to find what is wrong and what to fix first.
Crawlability, indexability, security headers, Core Web Vitals, JS rendering. Start here when pages are not getting indexed.
One URL, examined closely. Good for a page that should rank and does not.
E-E-A-T, depth, readability, thin content, and whether an AI engine could cite the page.
Detects what structured data exists, validates it, and generates what is missing.
Validates an existing sitemap or generates one, with quality gates against thin location pages.
Alt text, file size, formats, lazy loading, CLS, and image metadata.
Visibility inside AI answers rather than blue links: AI Overviews, ChatGPT, Perplexity, Copilot.
Decide what to build before building it.
Seed expansion, intent mapping, difficulty scoring. The input to everything in Research and planning.
Groups keywords by real SERP overlap, not text similarity, then designs the hub-and-spoke architecture.
A writer-ready brief: outline, word counts per section, competitor scoring, internal links.
Strategy and roadmap by industry, when you need the shape of the programme rather than a single fix.
Comparison and alternatives pages, which are usually the fastest commercial wins available.
Pages generated at scale from data, with the thin-content and index-bloat safeguards that make it survivable.
For businesses that rank in a place, not just on a topic.
Google Business Profile, NAP consistency, citations, reviews, map pack.
Geo-grid rank tracking and competitor radius mapping. Deeper than the local skill, and needs data extensions for the good parts.
Product catalogues and multi-language sites.
Product schema, Google Shopping and marketplace visibility, pricing gaps.
International targeting. Catches the return-tag and region-code errors that quietly break multi-language sites.
Real numbers from Google and the link graph.
First-party data: Search Console, PageSpeed, CrUX, GA4. Requires credentials; everything else here works without them.
Referring domains, anchor text, toxic links, competitor gaps, from free and paid sources.
Live SERP and keyword data. Needs the DataForSEO extension and costs money per query.
Keep what you built working, and find out when it stops.
Reads the SERP backwards to explain why a well-optimised page still does not rank. Usually the answer is page-type mismatch.
Baseline the SEO-critical elements, then diff after every deploy. Git for on-page SEO.
Finds decaying pages and says why, so you update rather than publish more.
Maps the link graph, finds orphans and weak hubs, and produces an add-this-link-here plan.
Redirect mapping and pre/post-launch checks for domain moves and replatforms.
Assembles prior analyses into something a stakeholder will actually read.
Get the work onto the site and into the index.
Turns an approved draft into a live page. Site-specific: it is wired to a particular repository and deploy pipeline.
Audits OG and social images and plans replacements. Generation needs the banana extension.
Surfaces stage-specific prompts from the FLOW evidence-led framework.
Three layers, and you only ever type the first one.
| Layer | What it is | You interact with it? |
|---|---|---|
| Skills | The 32 commands on this page. They decide what to do and in what order. | Yes — this is what you type. |
| Sub-agents | 18 specialists that a skill spawns in parallel for big jobs. A full audit runs up to 15 at once. | No — skills invoke them. |
| Scripts | 53 Python tools doing the actual fetching, parsing and API work in an isolated virtualenv. | Rarely — see Reference. |