Commands
| Command | What it does |
|---|
/seo sxo <url> [keyword] | Full SXO analysis. If the keyword is omitted, the skill detects it from the page title and H1 |
/seo sxo personas <url> | Persona-only scoring. Skips SERP analysis. Useful when you already know the target page type and just want UX-side feedback |
/seo sxo wireframe <url> | IST and SOLL wireframes with concrete placeholders based on the page, SERP consensus, and persona gaps |
Evidence and access boundaries
SERP evidence is a dated locale and device snapshot. DataForSEO requires the user's credentials, consumes paid credits, and should run only after a cost estimate and confirmation. The WebSearch fallback has lower precision. Sparse results, personalization, ads, JavaScript, and changing result features can reduce confidence. Query-fit conclusions and wireframes are hypotheses for human review and testing, not ranking forecasts.
How Page-Type Classification Works
The classifier uses a hybrid rubric of structural and content signals. No single signal decides a page type. The combination does.
- URL pattern:
/blog/ vs /products/ vs /vs/ vs /best-* are strong early signals
- H1 + title shape: 'How to X', 'X vs Y', 'Best X for Y', 'X review', 'Buy X' each map to different page types
- Heading hierarchy: tutorials have step-numbered H2s. Comparisons have feature-named H2s. Listicles have ranked-item H2s
- Schema markup: Product, Article, ItemList, VideoObject, and other currently supported types can reinforce classification. FAQPage and HowTo are excluded from expected Google SERP features
- Content blocks: presence of pricing tables, comparison matrices, step lists, embedded video, CTAs above fold
- Word count tier: an internal classification clue, not a Google threshold or a requirement to add words
Each top-10 result is classified with the same rubric. The most common category describes that snapshot, but it is not a permanent intent label or a ranking rule. If your page differs, the skill assigns an internal review severity and proposes options to investigate.
Cross-Skill References
SXO does not duplicate the rest of Claude SEO. It hands off when a specialist is the right next step:
- E-E-A-T gaps in persona scoring → /seo content for deep content audit
- Missing schema types → /seo schema for generation
- Local intent detected → /seo local for GBP analysis
- AI search angles → /seo geo for AI Overviews and citation readiness
Attribution
SXO was contributed by Florian Schmitz via the AI Marketing Hub Pro Hub Challenge. The original repo lives at github.com/tools-enerix/claude-sxo-skill. Integrated into Claude SEO v1.9.0 with Florian's permission. The skill code is MIT-licensed, like the rest of Claude SEO.
