SEO Skills for AI Coding Agents

Best for: A first-pass audit when no specialist dependency is required

Addy Osmani's seo skill is the default starting point for a focused SEO review. It gives an agent a practical checklist for examining common technical and on-page concerns and turning the review into concrete next steps.

It is guidance rather than a hosted crawler or a measurement system. Use it to structure an audit of the evidence you provide; it does not establish that a site will rank, be indexed, receive citations, or gain traffic.

• Default choice for a general SEO audit
• Technical and on-page review guidance
• No crawler or external service required by the skill
• Recommendations remain bounded by the evidence supplied

Best for: A guided diagnosis when you already use Corey's marketingskills suite

Corey Haines' seo-audit is a conditional choice for a guided diagnosis. It helps an agent clarify the site's context and SEO problem, then work through a structured technical and on-page audit from the information and evidence available.

Its boundary matters: this is an advisory audit skill within the existing marketingskills suite, not a replacement for a crawler, Search Console data, or the suite's separate skills for adjacent work. Treat its output as diagnosis and prioritization, not as proof of a future search result.

• Conditional choice for Corey's existing marketingskills suite
• Guided diagnosis before recommendations
• Uses supplied context and available evidence
• Not a crawler or a substitute for adjacent suite skills

Best for: Adding or correcting page-faithful JSON-LD and validating it

schema-markup-generator selects schema types, maps visible page content to their properties, generates JSON-LD, and gives implementation and validation guidance. It covers types including FAQPage, HowTo, Article, Product, LocalBusiness, and BreadcrumbList.

The workflow requires schema to match the visible page content and labels rich results as eligibility rather than a promise.

• Matches markup to the page's real content
• Uses JSON-LD and supported schema types
• Includes Rich Results Test and Schema.org validation
• Useful for reusable markup on dynamic pages

Best for: Planning page hierarchy, URLs, navigation, and internal links

site-structure-optimizer works in architecture mode for a whole-site hierarchy and in linking mode for an existing link graph. It outputs a URL map, hub-and-spoke plan, orphan disposition, and prioritized anchor plan as appropriate.

It keeps external backlinks and indexation diagnosis out of scope so its structure decisions can feed the correct follow-up workflow.

• Separates architecture and linking modes
• Maps hierarchy, URLs, navigation, and hub-and-spoke clusters
• Identifies orphans and anchor-text opportunities
• Produces a prioritized structure handoff

Best for: Diagnosing LCP, INP, or CLS with field and lab evidence

core-web-vitals starts with real-user signals and browser traces before changing code. It then narrows investigation to the measured LCP, INP, or CLS issue and calls for an equivalent follow-up measurement.

When only source is available, it labels findings as hypotheses rather than claiming a measured regression.

• Measures before optimizing
• Targets LCP, INP, and CLS separately
• Uses field data and browser traces when available
• Keeps lab evidence distinct from field outcomes

Best for: Planning templates and data that produce useful, non-thin pages

programmatic-seo first scores whether a page system has a viable pattern, page-specific value, data quality, intent match, and sustainable maintenance path. It stops rather than recommends a weak pSEO project.

For viable systems, it defines the data model, template rules, indexation gates, and kill-switch conditions for thin-content or index-bloat signals.

• Starts with demand and data validation
• Requires unique value on every generated page
• Plans internal links and indexation alongside templates
• Flags thin pages, cannibalization, and doorway-page risks

Best for: Checking discoverability and citations across AI search surfaces

seo-geo audits AI-search readiness through crawler access, rendering, citation structure, brand signals, and platform-specific visibility. Its guide anchors the work in Google's stated people-first SEO position rather than treating GEO as a separate ranking trick.

It records the source and confidence of findings, including when AI-search visibility cannot be established from the available evidence.

• Checks crawler access and server-rendered content
• Reviews passages, authority signals, and platform visibility
• Keeps Google guidance ahead of community claims
• Labels limitations and evidence sources in the report

Best for: Finding repeatable competitor wins, weaknesses, and market gaps

competitor-analysis compares competitor SEO and GEO strategies to surface repeatable wins, weak spots, and market gaps. It returns a prioritized competitor brief for the next research or implementation step.

Use it when the choice depends on observed competitors rather than an assumed content or search opportunity.

• Compares SEO and GEO strategies
• Finds repeatable wins and weaknesses
• Surfaces market gaps
• Returns a prioritized research brief

Best for: Inspecting referring domains, anchors, toxic-link signals, and link gaps

seo-backlinks identifies available sources before assessing a backlink profile, then reports data sources and confidence alongside referring domains, anchors, quality, toxic-link signals, top linked pages, and competitor gaps.

When the evidence is insufficient, it refuses to invent a health score and instead reports the missing coverage. It can use Common Crawl and verification checks without premium data, with optional Moz, Bing, or DataForSEO enrichment.

• Detects available data sources before analysis
• Labels evidence and confidence for each metric
• Includes anchor, quality, toxic-link, and gap checks
• Refuses misleading scores when data coverage is too thin