avav25/seo-review

<!-- ARCHITECTURAL NOTE: no `context: fork`. This skill spawns `Agent(seo-engineer)` for review steps. Subagents cannot spawn other subagents (per Anthropic docs), so this skill MUST run in the main thread to retain spawn capability. -->

SEO Review

Systematic SEO audit for public-facing pages. Checks crawlability, indexability, on-page SEO, structured data, Core Web Vitals, and AI search readiness. Applies Agent(seo-engineer) for all steps.

Works standalone or as a sub-workflow called by /docs, /feature-dev, or /bugfix.

1. Define Scope

Ask the user (or extract from parent workflow):

• Target pages: Specific URLs, page group, or entire site
• Environment: Local dev / staging / production
• Review type: Full audit / metadata only / structured data / performance / AI search
• Constraints: What must not change (URL structure, navigation, etc.)

2. Apply Roles

Apply Agent(seo-engineer) as primary role.

If implementation changes are needed, delegate to:

| Change Type | Delegate To | |---|---| | Frontend code (meta tags, JSON-LD, components) | Agent(frontend-engineer) | | Content / copy improvements | Agent(content-writer) | | Server config (redirects, headers, robots.txt) | Agent(devops-engineer) | | Architecture changes (URL structure, routing) | Agent(solution-architect) |

3. Baseline Audit

Run automated checks before making changes.

3a. Crawlability

Check the following (read project files or use browser/curl):

• robots.txt — allows important pages, Sitemap: directive present
• XML sitemap — includes all indexable pages, excludes non-indexable
• No orphaned pages (all pages reachable via internal links)
• Navigation uses crawlable <a href> links
• No crawl traps (infinite URL parameters, calendar widgets)

3b. Indexability

• No accidental noindex on public pages
• Canonical tags point to correct URLs
• No canonical conflicts (canonical vs redirects vs internal links)
• Pages return 200 status codes (check for soft 404s)
• Critical content rendered server-side (SSR/SSG), not client-only

3c. On-Page SEO

For each target page:

| Element | Check | |---|---| | Title tag | Unique, descriptive, matches page intent | | Meta description | Unique, compelling, relevant | | H1 | Exactly one per page, matches topic | | Heading hierarchy | Logical H1→H2→H3 | | Image alt text | Descriptive for all meaningful images | | Internal links | Present, descriptive anchor text | | Outbound links | Qualified with rel attributes where needed |

3d. Structured Data + E-E-A-T Author Signals

• JSON-LD present where applicable:
  • Organization / WebSite on homepage
  • Article / BlogPosting on blog posts
  • BreadcrumbList for navigation
  • FAQ where genuinely useful
  • Product, Review, HowTo where applicable
• Structured data matches visible page content (no misleading markup)
• Validate with Rich Results Test

E-E-A-T author signals (Experience, Expertise, Authoritativeness, Trustworthiness — see Google's helpful content guidance):

• Person schema on every blog post with sameAs linking to author's LinkedIn, Wikipedia, personal site, or X profile
• dateModified current — refresh evergreen content within rolling 6 months
• Author byline pages exist, are crawlable (not noindex), and link from each post
• Organization schema on author pages reinforces publisher trust

3e. Core Web Vitals + Mobile + Modern Transport

Mobile-first indexing is universal in 2026 — mobile checks fold into CWV.

| Metric | Target | Tool | |---|---|---| | LCP (Largest Contentful Paint) | < 2.5s | PageSpeed Insights | | INP (Interaction to Next Paint) | < 200ms | PageSpeed Insights | | CLS (Cumulative Layout Shift) | < 0.1 | PageSpeed Insights |

Mobile rendering:

• Responsive across viewports, no horizontal scroll
• Tap targets ≥ 48×48px, body font ≥ 16px
• No intrusive interstitials

Modern transport baseline (verify with curl -I --http3 or browser devtools):

• HTTP/3 (QUIC) enabled at the edge
• TLS 1.3 served (TLS 1.2 acceptable as fallback; 1.0/1.1 disabled)
• HSTS header set on production hostnames

3f. AI Search Readiness

• Pages crawlable and well-structured (SSR/SSG, semantic HTML)
• Content is clear, factual, and authoritative
• Structured data present and accurate
• llms.txt present — recommended for AI-search visibility as an active discovery signal for AI engines (format: llmstxt.org). Google still does not honor it, but ChatGPT, Claude, and Perplexity surface it.
• Pages indexed and snippet-eligible (prerequisite for AI features)

3g. AI Bot Accessibility

The most-broken 2026 SEO/GEO surface: AI crawlers silently 403'd by CDN/WAF rules. Verify robots.txt allows current AI bots and that the edge layer (Cloudflare, Akamai, Fastly) is not blocking them.

Required robots.txt allowlist:

User-agent: GPTBot
User-agent: ClaudeBot
User-agent: Perplexity-User
User-agent: Google-Extended
User-agent: OAI-SearchBot
User-agent: Applebot-Extended
User-agent: CCBot
User-agent: meta-externalagent
Allow: /

CDN/WAF check — Cloudflare "Bot Fight Mode" and the "AI Scrapers and Crawlers" managed rule are the most common silent blockers. Either disable the rule or add explicit allow rules per AI bot UA string.

Verify each bot returns 200, not 403:

curl -A "GPTBot" https://example.com/ -I
curl -A "ClaudeBot" https://example.com/ -I
curl -A "Perplexity-User" https://example.com/ -I
curl -A "Google-Extended" https://example.com/ -I

Bot UA references (sources change — re-check on audit):

• OpenAI GPTBot — platform.openai.com/docs/bots
• Anthropic ClaudeBot — docs.anthropic.com/en/docs/agents-and-tools/web-search-tool/bot-info
• Google-Extended — developers.google.com/search/docs/crawling-indexing/overview-google-crawlers
• Perplexity — docs.perplexity.ai/guides/bots

3h. GEO/AEO Audit

Apply @geo-writer skill to audit extractability for AI engines (ChatGPT, Claude, Perplexity, Gemini, Google AI Overviews):

• Schema coverage — Article or BlogPosting + Person with sameAs + Organization on blog pages; FAQPage where FAQ present; HowTo for tutorials; Product / Review on commercial pages. JSON-LD must mirror visible text (no marketing-only prose in schema).
• Answer-first structure — each H2/H3 opens with a self-contained 30-60 word answer. No buried ledes.
• Macro/meso/micro chunking — single H1, 5-9 H2 sections phrased as questions or topics, paragraphs 40-80 words, bold anchors every section.
• Entity clarity — canonical brand terms used verbatim, no pronoun drift, entity mentions in first 100 words.
• High-leverage formats — FAQ blocks (4-8 Q&A), comparison tables (2+ entities), HowTo steps (3-8 numbered actions) where natural.
• Freshness signals — dateModified present and recent; datePublished set; content updated within the pillar's freshness window.
• Citation-ready — stats/claims attributed to named sources with outbound links where appropriate.

The geo-content rule enforces these for public-facing text. Use pre-publish-checklist.md from @geo-writer for the full checklist.

4. Present Findings

Compile audit results:

## SEO Audit Report

### Pages Reviewed: [count]

### Critical Issues (blocking indexing/ranking)
- [ ] [issue] — [affected page] — [fix]

### Warnings (affecting quality/performance)
- [ ] [issue] — [affected page] — [fix]

### Opportunities (improvements)
- [ ] [opportunity] — [affected page] — [recommendation]

### Scores
- Crawlability: [OK / issues found]
- Indexability: [OK / issues found]
- On-Page SEO: [OK / issues found]
- Structured Data + E-E-A-T: [OK / missing / errors]
- Core Web Vitals + Mobile + Transport: [LCP: Xs, INP: Xms, CLS: X.XX, HTTP/3: y/n, TLS 1.3: y/n]
- AI Search Readiness: [OK / opportunities]
- AI Bot Accessibility: [200 per bot / 403 found on: bot list]
- GEO/AEO: [OK / opportunities]

Wait for user to review and approve which fixes to implement.

5. Implement Fixes

For approved fixes:

1. Metadata changes (title, description, OG tags) — implement directly or delegate to Agent(frontend-engineer)
2. Structured data (JSON-LD) — implement directly or delegate to Agent(frontend-engineer)
3. Content changes — delegate to Agent(content-writer)
4. Server config (redirects, headers) — delegate to Agent(devops-engineer)
5. Performance fixes (image optimization, lazy loading, critical CSS) — delegate to Agent(frontend-engineer)

Follow Agent(seo-engineer) standards for all changes. No black-hat techniques.

6. Verify

Re-run relevant checks from Step 3 for changed pages:

• [ ] Fixed issues no longer present
• [ ] No new issues introduced
• [ ] Structured data validates (Rich Results Test)
• [ ] Pages still indexable (no accidental noindex)
• [ ] Core Web Vitals within targets
• [ ] No broken links introduced

7. Summary

## SEO Review Summary
- **Scope**: [pages/environment reviewed]
- **Issues found**: [critical: X, warnings: X, opportunities: X]
- **Fixes applied**:
  - [fix 1] — [page]
  - [fix 2] — [page]
- **Delegated to**: [roles involved]
- **Verification**: [pass/fail per check]
- **Monitoring**: [what to watch — Search Console, PageSpeed, etc.]
- **Follow-ups**: [items for future attention]

Integration

• Roles: Agent(seo-engineer) (primary), Agent(frontend-engineer) (implementation), Agent(content-writer) (content fixes)
• Skills: @geo-writer (GEO/AEO audit — step 3h), @humanizer (applied when content fixes are needed)
• Rules: geo-content (enforces GEO structure and schema on public-facing text), humanize-content (enforces humanizer pass)
• Called by: /feature-dev (SEO-relevant features), /docs (public content)