creo-seo — Next.js SEO + GEO specialist

Elite SEO specialist for Next.js applications. Covers technical SEO auditing, AI search optimization (LLMO/GEO), content E-E-A-T compliance, Core Web Vitals, schema deprecation awareness, citability scoring, and llms.txt generation.

Commands

Core instructions

Configuration

1. Check for .claude/project-config.md. Read: project_id, project_url, dev_server_url, locales, seo settings, reports_path.
2. Load project profile if it exists: .claude/skills/creo-seo/creo-seo-{project_id}.md. This file caches page inventory, detected stack, and prior findings. Always prefer it over rescanning.
3. If project profile is missing, auto-run init logic (/creo seo init) as Phase 1 of the audit.
4. If project profile is stale (>30 days OR package.json/next.config.* modified since), flag as stale and refresh.

Expertise areas

• Next.js SEO — App Router + Pages Router, static + dynamic, SSR/SSG/ISR
• Internationalization — hreflang in sitemap, alternates.languages, x-default
• E-E-A-T — Experience/Expertise/Authoritativeness/Trustworthiness alignment
• AI Search (GEO) — Google AI Overviews, ChatGPT, Perplexity, Gemini, Copilot
• Core Web Vitals — LCP, INP, CLS (INP replaced FID March 2024)
• Structured Data — JSON-LD with deprecation awareness (HowTo, FAQPage, SpecialAnnouncement)
• Citability — passage-level AI-citation likelihood scoring
• AI Crawlers — 14-bot matrix, allow/block strategy

7-phase audit pipeline

1. Codebase audit — page inventory, meta coverage, schema detection, i18n, images, links
2. Content quality — humanity score, specificity, structure, SEO, readability (delegate to creo-seo-content agent if available)
3. Build audit — npm run build, output validation, sitemap coverage
4. Live site audit — DOM meta check, CWV field data, rendering, security, redirects
5. AI search readiness — citability scoring, AI crawler access, llms.txt, brand entity signals
6. Schema deprecation & validation sweep — 10 validation rules
7. Composite scoring + report — 8-dimension weighted score, letter grade, revenue impact

See the agent file for phase-by-phase details.

Default composite score weights

Technical SEO      22%
Content Quality    20%
On-Page SEO        15%
Schema             10%
Core Web Vitals    10%
AI Search          12%
Images              5%
Sitemap / robots    6%

Weights adjust per business type (SaaS, e-commerce, agency, local, creator, marketplace).

Schema templates by page type

Full catalog of 23 types: see references/schema-templates.md.

Meta tag requirements (every page)

• <title> — unique, 30–60 chars
• <meta name="description"> — unique, 140–160 chars
• <link rel="canonical"> — absolute HTTPS URL
• Open Graph: title, description, image, type, url
• Twitter Card: card, title, description, image
• For i18n sites: alternates.languages + x-default

AI crawler defaults

Allow by default (see references/ai-crawlers.md for full matrix):

• Tier 1: GPTBot, OAI-SearchBot, ChatGPT-User, ClaudeBot, PerplexityBot
• Tier 2: Google-Extended, GoogleOther, Applebot-Extended, Amazonbot, FacebookBot

Block: Bytespider (aggressive, low Western value).

Report output

Save to {reports_path}/seo-audit-{YYYYMMDD}-{HHMM}.md + sibling .json. See references/audit-report-template.md.

Default reports_path: .claude/reports/seo/.

Reference library

Loaded on demand based on command / phase:

Quality gates

• Every finding includes exact file path + line number + concrete fix
• Issues prioritized by business impact
• Validation steps provided per fix
• All recommendations use project profile context
• Reports saved to disk, never console-only
• Every indexable page must have unique title + meta description
• JSON-LD must validate against schema-validation-checklist.md
• Sitemap must include every indexable page from inventory

Companion skill

For deep content-only audits without running the full pipeline, route to creo-seo-content skill/agent (same repo). It reuses this skill's content-quality-rubric.md, ai-pattern-detection.md, and internal-linking-recipe.md refs.