| name | seo |
| description | SEO optimization. Meta tags, structured data, Core Web Vitals, sitemap, robots.txt, Open Graph.
|
SEO -- SEO Optimization & Technical Auditing
Activate When
/godmode:seo, "SEO audit", "meta tags"- "structured data", "schema markup", "Core Web Vitals"
- "sitemap", "robots.txt", "Open Graph"
- Pre-ship quality gate or search ranking drops
Workflow
Step 1: Technical Discovery
curl -sf https://example.com/sitemap.xml | head -5
curl -sf https://example.com/robots.txt
npx lighthouse https://example.com \
--only-categories=seo --output=json
SEO DISCOVERY:
Target: <URL / entire site>
Framework: <Next.js | Nuxt | SvelteKit | static>
Rendering: SSR | SSG | CSR | ISR | hybrid
Step 2: Meta Tag Audit
TITLE: 50-60 chars, unique per page,
primary keyword near beginning, brand at end
DESCRIPTION: 150-160 chars, unique, has CTA
CANONICAL: every page, self-referencing,
trailing slash consistent, no canonical to 404
MINIMUM META TAGS PER PAGE:
<title>, <meta description>, <link canonical>
OG: og:type, og:title, og:description,
og:image (1200x630), og:url, og:site_name
Twitter: twitter:card, twitter:title,
twitter:description, twitter:image
JSON-LD structured data
Step 3: Structured Data (JSON-LD)
Page type -> Schema.org type:
Homepage: Organization (name, url, logo, sameAs)
Blog posts: Article (headline, author, date)
Products: Product (name, price, availability)
FAQ: FAQPage (Question/Answer pairs)
Breadcrumbs: BreadcrumbList (position, name)
Search: WebSite + SearchAction
IF structured data present: validate with
Google Rich Results Test (0 errors required)
IF structured data missing: add for page type
WHEN data doesn't match visible content: fix
(mismatch = manual penalty risk)
Step 4: Sitemap & Robots.txt
SITEMAP:
Location: /sitemap.xml, XML format
Check: missing pages, dead URLs, non-canonical
Limits: 50MB / 50K URLs -> split
Reference in robots.txt
ROBOTS.TXT:
Allow important pages
Block admin/login/API/internal
Block duplicate params (?sort=, ?filter=)
NEVER block CSS/JS (engines need to render)
Step 5: Core Web Vitals
TARGETS: LCP <2.5s, INP <200ms, CLS <0.1
LCP FIXES: preload LCP image, responsive srcset,
WebP/AVIF, inline critical CSS, reduce TTFB,
defer non-critical JS, fetchpriority="high"
CLS FIXES: explicit width/height on images/videos,
aspect-ratio CSS, reserve space for ads/embeds,
preload fonts, font-display:optional/swap
INP FIXES: break long tasks (>50ms) with yield,
requestIdleCallback, web workers,
debounce inputs, minimize DOM (<1500 elements)
Step 6: Open Graph & Social
OG Image: min 1200x630, JPG/PNG <8MB,
absolute URL, publicly accessible
Test: Facebook Debugger, Twitter Card Validator
Step 7: Monitoring
Track weekly: organic traffic, impressions/clicks,
keyword rankings, crawl errors
Track per deploy: Lighthouse CI scores
Assert: performance >=0.9, seo >=0.9,
LCP <2500, CLS <0.1
Track monthly: CWV field data (CrUX)
Step 8: Report
SEO AUDIT -- {target}:
Meta: {status} ({N} pages)
Canonical: {status}, Sitemap: {status}
CWV: LCP {N}s, INP {N}ms, CLS {N}
Structured data: {N} types, valid: {yes|no}
Lighthouse SEO: {N}/100
Verdict: PASS (>=90) | NEEDS WORK (>=70) | FAIL
Commit: "seo: {target} -- fix {N} issues"
Key Behaviors
- Crawlability first. Robots, sitemap, canonicals.
- CWV affects ranking. Google ranking signals.
- JSON-LD earns rich results. Improves CTR.
- Meta tags are the sales pitch. Write for humans.
- Open Graph is not optional.
- SEO is ongoing. Audit every deploy.
- Never ask to continue. Loop autonomously.
HARD RULES
- EVERY page: unique title (50-60 chars) + description.
- EVERY page: canonical URL. No duplicate content.
- EVERY image: alt text. Decorative = alt="".
- NEVER client-render SEO-critical content.
- Structured data MUST match visible content.
- AUTO-GENERATE sitemap on deploy.
- EVERY page: single H1, logical heading hierarchy.
- CWV targets: LCP <2.5s, INP <200ms, CLS <0.1.
Auto-Detection
grep -r "next-seo\|react-helmet\|vue-meta" \
package.json 2>/dev/null
ls public/sitemap.xml public/robots.txt 2>/dev/null
grep -r "JsonLd\|json-ld\|application/ld+json" \
src/ --include="*.tsx" --include="*.vue" -l 2>/dev/null
TSV Logging
Append to .godmode/seo-results.tsv:
timestamp\tpage\taction\tmetric\tbefore\tafter\tstatus
Output Format
Print: SEO: Lighthouse {N}/100. CWV: LCP {N}s, INP {N}ms, CLS {N}. Verdict: {verdict}.
Keep/Discard Discipline
KEEP if: Lighthouse SEO >= previous AND CWV held
AND structured data validates
DISCARD if: SEO dropped OR CWV regressed
OR structured data errors
Stop Conditions
STOP when:
- Lighthouse SEO >= 95 AND all CWV good
- All pages have titles/descriptions/canonicals
- Valid structured data on all page types
- User requests stop OR max 10 iterations
