--- name: seo-auditor description: Find AI-search visibility gaps in any website, entities, schema, canonicals. version: 7 targets: [claude, codex, cursor, gemini, opencode] ---
SEO Auditor
Audit the provided URL or page HTML for AI-search and traditional SEO issues.
Run in this order
1. **Crawl basics** — title tag (50–60 chars), meta description (120–155 chars), canonical URL present and self-referencing, robots meta not accidentally blocking indexing.
2. **Heading structure** — single H1, logical H2/H3 hierarchy, no heading duplication across tabs or modals, heading text matches page intent.
3. **Entity coverage** — identify the primary entity (person, product, company, place). Check if entity name, description, and at minimum one supporting attribute appear in body copy. Flag thin entity coverage.
4. **Schema markup** — look for JSON-LD or microdata. For SaaS: SoftwareApplication or Product schema. For blog: Article schema with datePublished. For local: LocalBusiness. Flag missing or malformed markup.
5. **AI-search signals** — does the page answer the likely conversational query directly in the first 150 words? Are there clear question-and-answer patterns a retrieval system could extract? Flag pages that bury the answer.
6. **Core Web Vitals proxies** — flag render-blocking scripts in <head>, images without width/height attributes (CLS risk), and missing loading="lazy" on below-fold images.
Output format
Return a structured report:
PASS | issue title
WARN | issue title — one-line fix
FAIL | issue title — one-line fixThen a **Priority fixes** section: top 3 items by estimated impact.
Do not explain what SEO is. Just run the audit and report findings.