pipeline-seo-geo-aeo

Standalone SEO/GEO/AEO audit pipeline — 7 stages: technical SEO, structured data, E-E-A-T, AEO, GEO, architecture, and report. Runs against a URL or content files, produces a severity-tiered report wi

SKILL.md

GitHub

/pipeline-seo-geo-aeo — SEO/GEO/AEO Audit Pipeline

Audits existing content or live URLs for SEO, GEO, and AEO readiness. Does NOT create content — that's /pipeline-content's job. This pipeline checks what exists and reports what needs fixing.

Input

  • url (required if no files): live URL to audit
  • files (required if no url): content files to audit (markdown, HTML)
  • scope (optional): full (default) | technical-only | content-only | geo-only
  • brand (optional): brand name for entity authority checks

Pipeline State

Create tracker at .gstack/pipeline-seo-geo-aeo-{date}.md:

## Pipeline: SEO/GEO/AEO Audit
Started: {timestamp}
Target: {url or files}
Scope: {scope}

| # | Stage | Status | Score | Top Issue |
|---|-------|--------|-------|-----------|
| 1 | TECHNICAL | pending | —/10 | — |
| 2 | STRUCTURED DATA | pending | —/10 | — |
| 3 | EEAT | pending | —/10 | — |
| 4 | AEO | pending | —/10 | — |
| 5 | GEO | pending | —/10 | — |
| 6 | ARCHITECTURE | pending | —/10 | — |
| 7 | REPORT | pending | — | — |

Stage 1: TECHNICAL

Load seo-aeo-best-practices/references/technical-seo.md.

Check (URL mode):

  • Fetch page HTML (use browse/curl)
  • Title tag: present, 50-60 chars, keyword in title
  • Meta description: present, 150-160 chars
  • Canonical tag: present, correct URL
  • Open Graph: og:title, og:description, og:image all present
  • HTTPS: URL uses HTTPS
  • Fetch /robots.txt: check AI crawler entries (GPTBot, ClaudeBot, PerplexityBot)
  • Check for <link rel="sitemap"> or /sitemap.xml

Check (file mode):

  • Verify metadata frontmatter or HTML <head> section
  • Check for canonical, OG tags in template/layout

Gate: Pass if no Critical issues. Missing canonical = High. Missing meta title = Critical.


Stage 2: STRUCTURED DATA

Load seo-aeo-best-practices/references/structured-data.md.

Check:

  • Extract all <script type="application/ld+json"> blocks
  • Identify schema types present
  • For each type: validate required fields against reference templates
  • Flag missing required fields, date format errors, mismatched content
  • Flag schema types that SHOULD be present but aren't:
    • Blog post without Article schema = High
    • FAQ section without FAQPage schema = Medium
    • Homepage without Organization schema = High

Gate: Pass if no Critical issues. Missing required field on existing schema = Critical. Expected schema entirely absent = High.


Stage 3: E-E-A-T (parallel with 4, 5)

Load seo-aeo-best-practices/references/eeat-principles.md.

Check:

  • Author attribution visible on page
  • Author bio page linked (check for /about/, /team/, /author/ links)
  • Publication/update date visible
  • Sources cited for factual claims (look for inline citations, references section)
  • YMYL topic detection (health/finance/legal keywords) — if detected, flag missing credentials as Critical
  • Person schema on author page (if accessible)
  • Organization schema with sameAs on homepage

Gate: Pass if no Critical issues. YMYL without credentials = always Critical.


Stage 4: AEO (parallel with 3, 5)

Load seo-aeo-best-practices/references/aeo-optimization.md.

Check:

  • Primary question answered in first 100 words
  • H2 headings present — are they question-phrased or topic-clear?
  • FAQPage schema present if FAQ section detected (cross-reference Stage 2)
  • Content length vs. intent category (flag padding or thinness)
  • Extraction zone present for long content (>1,500 words): TL;DR, summary, key takeaways

Gate: Pass if no Critical issues. No answer in first 100 words = High.


Stage 5: GEO (parallel with 3, 4)

Load seo-aeo-best-practices/references/geo-optimization.md.

Check:

  • BLUF structure: does each major section open with a direct statement?
  • Quantitative claims: count data points with sources
  • Organization schema with sameAs (cross-reference Stage 2)
  • AI crawler access: cross-reference Stage 1 robots.txt — flag if GPTBot/ClaudeBot/PerplexityBot blocked
  • Original data or third-party citations present
  • Entity authority: check if sameAs includes Wikipedia/Wikidata URLs

Note: Full GEO measurement (AI mention tracking) cannot be automated in a single audit. Flag as "Recommended: set up GEO monitoring" with tool suggestions.

Gate: Pass if AI crawlers not blocked and entity schema signals present. Everything else = Medium or Low.


Stage 6: ARCHITECTURE (skip if URL-only with no site access)

Load seo-aeo-best-practices/references/content-architecture.md.

Check:

  • Internal links: count outbound internal links on page, flag if zero
  • URL structure: check for length, date inclusion, parameter sprawl
  • Heading hierarchy: H1 → H2 → H3 (no skipped levels)
  • Content freshness: dateModified in schema vs. visible date

Gate: Skip gracefully if no site-wide access. Flag what can be checked on a single page.


Stage 7: REPORT

Load seo-aeo-best-practices/references/integration-checklist.md for scoring calibration.

Output format:

## SEO/GEO/AEO Audit Report

**Target:** {url or files}
**Date:** {YYYY-MM-DD}
**Scope:** {scope}

---

### Composite Score

| Area | Score | Grade | Top Issue |
|------|-------|-------|-----------|
| Technical SEO | X/10 | A-F | {one-liner} |
| Structured Data | X/10 | A-F | {one-liner} |
| E-E-A-T | X/10 | A-F | {one-liner} |
| AEO | X/10 | A-F | {one-liner} |
| GEO | X/10 | A-F | {one-liner} |
| Architecture | X/10 or SKIPPED | A-F | {one-liner} |

**Overall: X/60 (X/50 if Architecture skipped) — Grade: {letter}**

Grade scale: A (50-60) | B (40-49) | C (30-39) | D (20-29) | F (<20)

---

### Critical Issues (fix before publishing)
{severity-tiered findings}

### High Issues (fix within 1 week)
{findings}

### Medium Issues (address in next content update)
{findings}

### Low / Recommended Monitoring
{findings}

---

### Quick Wins (high impact, low effort — fix in under 30 minutes)
1. {item}
2. {item}
3. {item}

Update pipeline tracker with final scores, then present report to user.


Scope Shortcuts

Scope Stages run
full All 7 stages
technical-only Stage 1 + 2 + 7
content-only Stage 3 + 4 + 5 + 7
geo-only Stage 5 + 7

Integration with pipeline-content

When called from /pipeline-content as Stage 4b (optional deep SEO audit after content-critique passes):

  • Runs content-only scope by default (stages 3-5 + 7)
  • The audit supplements content-critique's dimension 5 score with itemized findings
  • If composite content score < 30/50: flag as blocker, recommend specific fixes before publishing