creo-seo git:20260817.079b5c9A

Immutable. This exact content is served forever at /api/v1/blob/c8602963f62e808c.

---
name: creo-seo
description: >
  Next.js SEO + GEO specialist. Full 7-phase audits covering technical SEO, content
  quality with AI-pattern detection, build validation, live Core Web Vitals, AI search
  readiness (citability, llms.txt, 14 AI crawlers), schema deprecation sweep, and
  composite scoring. Auto-builds per-project cached profile for faster repeat audits.
  Also runs the ongoing SEO operation: closed-loop GSC autofix with allowlist and
  ledger, coverage triage, semantic-core prioritization, weekly rank + LLM-citability
  monitoring, freshness guards, and off-page authority planning.
  Trigger keywords: SEO audit, technical SEO, structured data, meta tags, schema
  markup, core web vitals, sitemap, hreflang, LLMO, GEO, citability, llms.txt,
  AI crawlers, Search Console, GSC, indexation, coverage, semantic core, rank
  tracking.
allowed-tools:
  - Read
  - Grep
  - Glob
  - Bash
  - WebFetch
  - Agent
---

# 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

| Command | What it does |
|---------|-------------|
| `/creo seo onboard <url>` | **Main lifecycle command** — audit, implementation plan, guided GSC/Bing configuration, monitoring + autofix scaffold, standing cadence. See `references/seo-onboarding.md`. |
| `/creo seo plan <url>` | Analyze + produce a prioritized implementation plan (Problem/Fix/Verify, DEV/YOU tags, phased forecast) without the setup stages |
| `/creo seo init` | Scan codebase and write cached project profile to `.claude/skills/creo-seo/creo-seo-{project_id}.md`. Run this first. |
| `/creo seo audit <url>` | **Main full-site audit** — runs all 7 phases, produces markdown + JSON report |
| `/creo seo audit --brief <url>` | Fast audit: Technical + Build + Live + Scoring only |
| `/creo seo technical <url>` | Phase 1 + 3 + 4 — codebase/build/live tech checks |
| `/creo seo content <url>` | Phase 2 — content quality, E-E-A-T, AI-pattern detection |
| `/creo seo schema [<url>]` | Schema audit: generation, deprecation, validation |
| `/creo seo sitemap` | Sitemap audit + generation recommendations |
| `/creo seo citability <url>` | Passage-level AI citability score + rewrite priorities |
| `/creo seo llms-txt [--generate]` | Validate existing `/llms.txt` or generate new one |
| `/creo seo crawlers [--fix]` | Audit robots.txt for 14 AI crawlers; optionally rewrite |
| `/creo seo compare <old.md> <new.md>` | Delta between two prior audits |
| `/creo seo autofix [--dry-run]` | Closed-loop GSC remediation: detect, allowlisted fix, verify, notify, ledger (needs gsc-analyzer extension) |
| `/creo seo triage <bucket>` | Triage a GSC coverage bucket: expected vs concern, direction over level |
| `/creo seo semantic-core` | Build/refresh the focused query core with noise filtering and P0-P3 priorities |
| `/creo seo freshness` | Audit freshness signals: lastmod/dateModified divergence, year-staleness |
| `/creo seo offpage` | Off-page authority plan: directories, LLM citation slots, trust-killer sweep |
| `/creo seo weekly [--setup]` | Read the latest weekly snapshot, or scaffold the weekly monitoring workflow |
| `/creo seo page-rules` | Generate a project-scoped `.claude/rules/seo-page-rule.md` for page-creation-time enforcement |

## 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

| Page type | Primary | Complementary |
|-----------|---------|---------------|
| Homepage | Organization + WebSite | sameAs, SearchAction |
| Product | Product | Offer, AggregateRating, Review |
| Pricing | SoftwareApplication / Offer catalog | — |
| Blog post | BlogPosting | Person, BreadcrumbList |
| Local | LocalBusiness (+subtype) | PostalAddress, GeoCoordinates |
| Event | Event | Offer, Place |
| Course | Course | CourseInstance |
| Job | JobPosting | — |
| Video | VideoObject | — |

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:

| File | Purpose |
|------|---------|
| `references/project-profile-template.md` | What `/creo seo init` writes |
| `references/audit-report-template.md` | What `/creo seo audit` writes |
| `references/scoring-rubric.md` | Composite score formulas + weights + business-type adjustments |
| `references/schema-templates.md` | 23 JSON-LD templates + deprecation calendar + Next.js helpers |
| `references/schema-validation-checklist.md` | 10 pre-publish schema rules |
| `references/sitemap-patterns.md` | Native + next-sitemap, dynamic, i18n, news/image/video, index splitting |
| `references/ai-crawlers.md` | 14-bot matrix + robots.txt templates + Next.js `app/robots.ts` |
| `references/llms-txt-generator.md` | Spec, sections, validation, scoring, Next.js generator |
| `references/geo-citability.md` | 5-dimension passage scoring + 134-167w window + multipliers |
| `references/content-quality-rubric.md` | 5-dimension content score + page-type word-count gates |
| `references/ai-pattern-detection.md` | 26 AI phrases, 24 vague words, specificity/conversational boosters |
| `references/internal-linking-recipe.md` | Link-per-page gates, anchor strategy, cluster model, Next.js patterns |
| `references/gsc-autofix-loop.md` | Closed-loop GSC remediation: fix allowlist, anti-thrash ledger, run sequence |
| `references/coverage-triage.md` | Reading GSC indexing buckets: expected vs concern, direction over level |
| `references/deploy-verification.md` | Edge-vs-origin checks, redirect assertions, canaries, validation windows |
| `references/semantic-core.md` | Focused query core: noise filtering, P0-P3, opportunity buckets, trend labels |
| `references/freshness-signals.md` | git-mtime as single source of truth, year policy, staleness CI guard |
| `references/page-seo-rules.md` | Page-creation-time SEO rules (metadata vs JSON-LD, route enums, checklists) |
| `references/seo-program-conventions.md` | DEV/YOU tags, Problem/Fix/Verify, changelogs, repeat-audit reports |
| `references/indexation-runbook.md` | Hypothesis-branch template for missing/unindexed route families |
| `references/offpage-authority.md` | Directory playbook, LLM citation slots, trust-killer sweep, Bing coverage |
| `references/seo-onboarding.md` | `/creo seo onboard` lifecycle: audit -> plan -> GSC setup -> monitoring cadence |

## 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

## Operations mode (Search Console lifecycle)

Beyond point-in-time audits, this skill runs an ongoing SEO operation when the
`gsc-analyzer` extension is installed (`extensions/gsc-analyzer/` — scripts,
templates, GSC API reference, UI automation):

- **Autofix** (`/creo seo autofix`) — the weekly closed loop per
  `references/gsc-autofix-loop.md`: detect via budgeted URL inspection, fix
  only allowlisted issue classes, verify live at origin AND edge
  (`references/deploy-verification.md`), notify Google, track everything in
  an anti-thrash ledger. Judgment-shaped issues become GitHub issues, never
  edits.
- **Triage** (`/creo seo triage`) — before fixing anything from a coverage
  report, apply `references/coverage-triage.md`: many buckets are expected
  canonicalization behavior, most 404/redirect rows self-heal, direction
  matters more than level.
- **Monitoring** (`/creo seo weekly`) — weekly rank-history snapshot on the
  focused core, optional Bing merge and LLM-citability tracking, year-
  staleness check, delivered as a PR (extension `templates/seo-weekly.yml`).
- **Semantic core** (`/creo seo semantic-core`) — per
  `references/semantic-core.md`, with the per-project taxonomy in
  `seo-site-config.json` (extension template).

Without the extension, the operational commands still work in degraded mode:
the skill reads exported CSVs, runs curl-based verification, and produces the
same reports manually.

## 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.