Agent Skills: SEO/GEO Orchestrator

SEO/SEA/GEO 2026 complete methodology for organic, paid, and AI search optimization.

UncategorizedID: fusengine/agents/seo

Install this agent skill to your local

pnpm dlx add-skill https://github.com/fusengine/agents/tree/HEAD/plugins/seo/skills/seo

Skill Files

Browse the full folder contents for seo.

Download Skill

Loading file tree…

plugins/seo/skills/seo/SKILL.md

Skill Metadata

Name
seo
Description
Use when running SEO, GEO, schema, Core Web Vitals, sitemap, hreflang, E-E-A-T, AI Overviews, technical SEO, or structured data tasks. Covers full-site audits, single-page analysis, schema markup, content quality, AI search optimization, local SEO, sitemap/robots, internal linking, semantic clustering, and search experience.

SEO/GEO Orchestrator

Invocation: /seo $1 $2 where $1 is the command and $2 is the URL or local path.

Comprehensive SEO + GEO 2026 analysis across all industries (SaaS, local services, e-commerce, publishers, agencies). Orchestrates 19 sub-skills and 8 parallel sub-agents. Local-first, zero third-party APIs.

Quick Reference

| Command | What it does | Sub-skill | |---------|--------------|-----------| | /seo audit <url> | Full audit with parallel subagent delegation | seo-audit | | /seo page <url\|path> | Deep single-page analysis | seo-page | | /seo technical <url> | Technical SEO (robots, CWV, crawlability) | seo-technical | | /seo schema <url\|path> | Detect/validate/generate JSON-LD | seo-schema | | /seo content <url> | E-E-A-T + anti-cannibalization | seo-content | | /seo images <url> | Image SEO: alt text, formats, lazy loading | seo-images | | /seo sitemap <url\|generate> | Analyze or generate XML sitemaps | seo-sitemap | | /seo hreflang <url> | i18n/multilingual audit and generation | seo-hreflang | | /seo internal-linking <url> | Pillar/cluster, anchor optimization | seo-internal-linking | | /seo cluster <keyword> | SERP-based semantic clustering | seo-cluster | | /seo brief <topic> | Generate detailed content brief | seo-content-brief | | /seo plan <type> | Strategic planning (saas/local/ecommerce/publisher) | seo-plan | | /seo sxo <url> | Search Experience Optimization | seo-sxo | | /seo snippets <url> | Position 0 + AI Overviews input | seo-featured-snippets | | /seo local <url> | Local SEO (GBP, NAP, citations, map pack) | seo-local | | /seo geo <url> | AI Overviews/ChatGPT/Perplexity/Claude readiness | seo-geo | | /seo ecommerce <url> | Product schema, faceted nav, marketplace | seo-ecommerce | | /seo video <url> | VideoObject, YouTube, transcripts | seo-video | | /seo redirects <url> | 301/302, migration, chains | seo-redirects | | /seo drift <url> | SEO diff vs previous git HEAD | (uses diff-seo.ts) |

Orchestration Logic — /seo audit

When user invokes /seo audit <url>:

  1. Detect business type from homepage (SaaS, local, ecommerce, publisher, agency)
  2. Spawn sub-agents in parallel (single message, multiple Agent calls):
    • seo-technical (robots, sitemap, CWV, hreflang)
    • seo-schema (JSON-LD detect/validate)
    • seo-content (E-E-A-T, anti-cannibalization)
    • seo-geo (AI Overviews readiness)
    • seo-images (alt, formats, optim)
    • seo-sitemap (sitemap.xml + robots.txt)
  3. If local business detected → also spawn seo-local
  4. If content strategy signals (blog, pillar pages) → also spawn seo-cluster
  5. Collect results, generate unified report with SEO Health Score (0-100)
  6. Create prioritized action plan (Critical → High → Medium → Low)

Workflow (7 phases — for content-creation flows)

PHASE 1: ANALYZE   → Extract content, detect intent
PHASE 2: RESEARCH  → SERP analysis, 2026 trends, AI platforms
PHASE 3: KEYWORDS  → Extract with anti-cannibalization
PHASE 4: STRUCTURE → Meta, OG, Twitter Cards, Hn, schema, alt
PHASE 5: CONTENT   → Write SEO+GEO optimized content
PHASE 6: VALIDATE  → SEO + GEO compliance checklists
PHASE 7: DRIFT     → Compare to git HEAD (diff-seo.ts)

Reference Library (legacy 10 categories)

references/
├── 01-seo-foundations/   → research workflow, SEO vs GEO
├── 02-onpage-seo/        → meta tags, OG, Twitter, headers, alt
├── 03-schema-org/        → 9 schema types (templates ready)
├── 04-geo-2026/          → AI platforms, citations, zero-click
├── 05-technical-seo/     → CWV, mobile-first, crawlability
├── 06-content-strategy/  → E-E-A-T, anti-cannibalization, keywords
├── 07-sea-google-ads/    → Quality Score, landing pages (optional)
├── 08-measurement/       → GEO tracking, Share of Model
├── 09-checklists/        → pre-publication, technical audit
└── 10-local-seo/         → GBP, NAP, citations, Local Pack

Scripts Available (scripts/)

| Script | Purpose | |--------|---------| | parse-meta.ts | Extract title/description/OG/canonical via cheerio | | validate-schema.ts | Validate JSON-LD offline against schema.org dumps | | check-cwv.ts | Lighthouse CLI local wrapper | | parse-sitemap.ts | Validate sitemap.xml structure | | parse-robots.ts | Validate robots.txt directives | | parse-hreflang.ts | Validate hreflang tags | | diff-seo.ts | SEO drift via git diff | | geo-score.ts | LLM-readiness scoring (0-100) |

Differentiation

  • Local-first: zero Google APIs, zero DataForSEO/Moz, zero Python
  • Framework-native: delegates implementation to fuse-astro, fuse-nextjs, fuse-laravel
  • GEO-first: AI Overviews/ChatGPT/Perplexity as primary target, not afterthought
  • TS/Bun stack: all scripts in TypeScript, executed by Bun

Industry Detection

Detect from homepage signals:

  • SaaS: /pricing, /features, /integrations, "free trial", "sign up"
  • Local: address, phone, opening hours, GBP link, NAP
  • E-commerce: /products, /cart, /checkout, product schema
  • Publisher: /articles, NewsArticle schema, byline, dateline
  • Agency: /services, /case-studies, /clients, testimonials

Quality Gates

  • 30+ location pages → warning
  • 50+ location pages → hard stop, require manual audit
  • 100+ programmatic pages → warning
  • 500+ programmatic pages → hard stop
  • Thin content detection per page type
  • Doorway page prevention

First Invocation — Activate Hook

When user invokes /seo for the first time on a project:

  1. Check if .fuse-seo marker exists at project root (or any parent up to repo root)
  2. If absent, ask the user before any other work:

    "Activate the fuse-seo hook on this project? It will validate meta/schema/OG on every Write/Edit of HTML-like files (.html, .astro, .tsx, .vue, .blade.php) via hooks/validate-seo.ts."

  3. If user confirms → create empty file .fuse-seo at project root via touch .fuse-seo (or Write tool with empty content)
  4. If user declines → proceed without creating the marker (hook stays dormant on this project)
  5. Do not ask again in the same session — the marker presence is the persistent answer

This is opt-in by design: never auto-create without explicit user consent.

After Any Analysis

  1. Generate markdown report in .fuse-seo/reports/<date>-<command>.md
  2. Offer drift baseline capture: /seo drift <url> for future comparisons
  3. Suggest framework-specific implementation via delegation:
    • Astro project → fuse-astro:astro-seo
    • Next.js → fuse-nextjs:nextjs-stack
    • Laravel → fuse-laravel:laravel-blade