菜单
Audit websites for SEO issues and optimize content for search engine visibility.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Design static ad creatives for social media and display advertising campaigns.
Source and evaluate candidates with job analysis, CV screening, and pipeline tracking.
Find relevant companies and leads for B2B sales with ICP definition and qualification frameworks.
Draft emails, manage calendars, prepare agendas, and organize productivity.
Create brand identity kits — logos, color palettes, typography, naming, and style guides.
Conduct thorough, multi-source research with structured reports and source scoring.
| name | seo-auditor |
| description | Audit websites for SEO issues and optimize content for search engine visibility. |
Audit websites for technical SEO issues, analyze on-page optimization, and provide actionable recommendations to improve search engine visibility and rankings.
If the active task or system prompt says you are running an SEO scan, or the ReportSeoScanComplete tool is available, this skill is reference material only. In that mode:
seo-audit.md, seo-audit-report.md, or a chat-only audit summary..local/new_seo_issues/<group>/..local/existing_seo_issues/ only when preserving, dismissing, or superseding prior findings.ReportSeoScanComplete after the issue files represent the final scan result.User wants an SEO audit of their website
User asks how to improve search rankings
User wants to optimize content for specific keywords
User needs meta tag, title, or description improvements
User wants to compare their SEO against competitors
Paid advertising strategy (use ad-creative skill)
Social media content creation (use content-machine skill)
General competitive analysis without SEO focus (use competitive-analysis)
Building pages at scale for SEO (use programmatic-seo skill)
Before anything else, determine whether the site is a React/Vue/Angular SPA or server-side rendered. This is the single most important distinction in a modern SEO audit.
SPA (React, Vue, Angular):Googlebot cannot reliably execute JavaScript. Any page that requires the JS bundle to render iseffectively invisible to search engines. The <title>and<meta>inindex.html are all Google sees.
SSR (Express, Next.js, Nuxt, SvelteKit): Full HTML is returned to the crawler. Everything is indexable.
Hybrid: Many apps are SPA for authenticated pages but SSR for public/marketing pages. Identify which routes are which.
How to detect: curl -s https://domain.com/some-page | grep "<h1" — if no H1 is in the curl output but is visible in the browser, the page is SPA-rendered and invisible to Googlebot.
For SPA+SSR hybrids: Audit only the SSR pages deeply. Note which routes are SPA (and therefore invisible) without alarming the user — authenticated dashboards being SPA is expected and fine.
Crawlability & Indexation (can Google find and index it?)
Technical Foundations (is the site fast and functional?)
On-Page Optimization (is content optimized?)
Content Quality (does it deserve to rank?)
Authority & Links (does it have credibility?)
Check for unintentional blocks
Verify important pages allowed
Check sitemap reference
⚠️ Verify actual content, not just HTTP status. A React SPA returns 200 for any URL — including /robots.txtif no Express route handles it. Alwayscurl -s /robots.txt | head -5and confirm the output isUser-agent:plain text, not<!DOCTYPE html>.
Exists and accessible
Contains only canonical, indexable URLs
Updated regularly
⚠️ Same SPA trap applies to /sitemap.xml.curl /sitemap.xmland confirm the first line is<?xml version=. If it returns<!DOCTYPE html>, the sitemap route is missing and the SPA is serving as fallback.
For Express/Node backends with pSEO content: implement sitemap as a dynamic Express route that maps over data arrays (e.g., BANK_GUIDES,GLOSSARY_TERMS). This way, adding new entries to the arrays automatically updates the sitemap — no manual maintenance needed.
Important pages within 3 clicks of homepage
Logical hierarchy
No orphan pages (pages with no internal links)
Noindex tags on important pages
Canonicals pointing wrong direction
Redirect chains/loops
Soft 404s
Duplicate content without canonicals
All pages have canonical tags
HTTP → HTTPS canonicals
www vs. non-www consistency
Trailing slash consistency
LCP (Largest Contentful Paint): < 2.5s
INP (Interaction to Next Paint): < 200ms — replaced FID in 2025 as the responsiveness metric
CLS (Cumulative Layout Shift): < 0.1
Server response time (TTFB)
Image optimization and modern formats (WebP)
JavaScript execution and bundle size
CSS delivery
Caching headers and CDN usage
Font loading strategy
Check client/index.html(the actual source file, not just what Googlebot sees) for bloated Google Fonts requests. A single<link href="fonts.googleapis.com/css2?family=Inter&family=Roboto&family=Poppins&..."> loading 10+ font families is a render-blocking LCP killer.
<link rel="preconnect" href="https://fonts.googleapis.com">
<link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
<link rel="preload" as="style" href="https://fonts.googleapis.com/css2?family=Inter:wght@400;700&display=swap">
<link rel="stylesheet" href="https://fonts.googleapis.com/css2?family=Inter:wght@400;700&display=swap" media="print" onload="this.media='all'">
<noscript><link rel="stylesheet" href="https://fonts.googleapis.com/css2?family=Inter:wght@400;700&display=swap"></noscript>
Apply this pattern to both client/index.html (SPA) and any SSR shared shell functions.
Responsive design (not separate m. site)
Tap target sizes
Viewport configured
No horizontal scroll
Mobile-first indexing readiness
HTTPS across entire site
Valid SSL certificate
No mixed content
HSTS header
Readable, descriptive URLs
Keywords where natural
Consistent structure (lowercase, hyphen-separated)
No unnecessary parameters
Note: Google now excludes pages returning non-200 status codes (4xx, 5xx) from the rendering queue entirely — critical for SPAs.
Unique per page, 50-65 characters
Primary keyword near beginning
Compelling and click-worthy
Brand name at end with |separator (not-or—)
Common issues: duplicates, too long/short, keyword stuffing, missing
Watch for Unicode symbols in H1s (e.g., ↔,→,©). These look odd in SERPs and can be keyword-unfriendly. Use plain text equivalents.
Unique per page, 150-160 characters
Includes primary keyword
Clear value proposition with CTA
Common issues: duplicates, auto-generated, no compelling reason to click
One H1 per page containing primary keyword
Logical hierarchy (H1 → H2 → H3, no skipping)
Headings describe content, not used just for styling
Keyword in first 100 words
Related keywords naturally used
Sufficient depth for topic
Answers search intent
Better than current top-ranking competitors
Descriptive file names and alt text
Compressed file sizes, modern formats (WebP)
Lazy loading, responsive images
Important pages well-linked with descriptive anchor text
No broken internal links
No orphan pages
Clear primary keyword target
Title, H1, URL aligned with keyword
Content satisfies search intent
Not competing with other pages (cannibalization)
Experience: First-hand experience demonstrated, original insights/data, real examples
Expertise: Author credentials visible, accurate and detailed information, properly sourced claims
Authoritativeness: Recognized in the space, cited by others, industry credentials
Trustworthiness: Accurate information, transparent about business, contact info available, privacy policy, HTTPS
Every SSR page should have the appropriate schema type. Use an array-based approach in your shared shell function so each schema is its own <script type="application/ld+json">block — never concatenate JSON strings and inject them inside an existing<script> tag, as this creates malformed HTML with nested script tags.
// Accept an array of schema JSON strings
schemaJsons?: string[];
// Render each as its own script block
allSchemas.map(s => `<script type="application/ld+json">${s}</script>`).join("\n")
WebSite+Organization @graph — entity identity for Google's Knowledge Graph, sitelinks signalsBreadcrumbList — add to every page that has visible breadcrumb navigation, even if the page already has another schema type (HowTo, DefinedTerm, etc.)HowTo — step-by-step guides
FAQPage — Q&A content (excellent for AI Overviews / featured snippets)
DefinedTerm + DefinedTermSet — glossary terms
ItemList — hub/index pages listing multiple items
Article / BlogPosting — editorial content (include datePublished,dateModified)
Always include datePublishedanddateModified in schema — freshness signals help maintain rankings on competitive queries.
OG Image requirement: og:imagerequires an actual hosted image URL (1200×630px). Base64-embedded images cannot be OG images. If no hosted image exists, note it as a blocker for social sharing and skip theog:image tag rather than pointing to a non-existent URL.
<meta property="og:title" content="..." />
<meta property="og:description" content="..." />
<meta property="og:url" content="..." />
<meta property="og:type" content="website" />
<meta property="og:locale" content="..." /> <!-- e.g., es_DO, en_US -->
<meta property="og:site_name" content="..." />
<meta name="twitter:card" content="summary_large_image" />
<meta name="twitter:title" content="..." />
<meta name="twitter:description" content="..." />
<meta name="twitter:site" content="@handle" />
Schema Markup Detection Warning: webFetchandcurl cannot reliably detect structured data — many CMS plugins inject JSON-LD via client-side JavaScript. Never report "no schema found" based solely on webFetch. Recommend using Google Rich Results Test or browser DevTools for accurate schema verification. For SSR pages, curl is reliable.
Review robots.txt to differentiate between beneficial retrieval agents (OAI-SearchBot, Googlebot) and non-beneficial training scrapers
Use structured data (schema.org) as the language of LLMs
Use "BLUF" (Bottom Line Up Front) formatting to help content get cited in AI Overviews
Search for target keywords and analyze top-ranking pages
Identify content gaps and opportunities
Compare meta tags, content depth, structure, and E-E-A-T signals
When a site uses a shared HTML shell function (common in Express/Node SSR setups), a single change to that function fixes all pages simultaneously. This is the highest-leverage opportunity in an SSR SEO audit.
og:locale,og:site_name — often missing
Twitter Card tags — almost always missing
<meta name="theme-color"> — small trust/UX signal
Async font loading — often render-blocking
WebSite+Organization JSON-LD — missing from most sites
Canonical tag structure — verify it's using the correct canonical per page
BreadcrumbList schema — specific to each page's breadcrumb pathdatePublished/dateModified — specific to each content type's schemaSaaS/Product Sites: Product pages lack content depth, blog not integrated with product pages, missing comparison/alternative pages, thin feature pages
robots.txt and sitemap.xml not handled by Express — SPA returns 200 with HTML instead of proper plain text / XML
Massive font bundles in index.html affecting landing page LCP
Missing OG/Twitter tags in index.html (since that's all social crawlers see for SPA pages)
SSR pages missing BreadcrumbList schema despite having breadcrumb HTML
Shared shell function missing Twitter Card, og:locale, og:site_name — fixing once would help all pages
E-commerce: Thin category pages, duplicate product descriptions, missing product schema, faceted navigation creating duplicates
Content/Blog Sites: Outdated content not refreshed, keyword cannibalization, no topical clustering, poor internal linking
Local Business: Inconsistent NAP, missing local schema, no Google Business Profile optimization
# SEO Audit Report: [Website]
## Executive Summary
- Overall health assessment
- Top 3-5 priority issues
- Quick wins identified
## Critical Issues (Fix Immediately)
| Issue | Page | Impact | Evidence | Fix |
|-------|------|--------|----------|-----|
## High-Impact Improvements
| Issue | Page | Impact | Evidence | Fix |
|-------|------|--------|----------|-----|
## Quick Wins
| Opportunity | Page | Potential Impact |
|------------|------|-----------------|
## Page-by-Page Analysis
### [Page URL]
- **Title**: Current | Recommended
- **Meta Description**: Current | Recommended
- **H1**: Current | Recommended
- **Content Score**: X/10
- **Issues**: [list]
## Prioritized Action Plan
1. Critical fixes (blocking indexation/ranking)
2. High-impact improvements (SSR shell function — fix once, applies to all pages)
3. Quick wins (easy, immediate benefit)
4. Long-term recommendations (OG image creation, Privacy/Terms SSR, etc.)
Free: Google Search Console, Google PageSpeed Insights, Rich Results Test (use for schema validation — it renders JavaScript), Mobile-Friendly Test, Schema Validator
Paid (if available): Screaming Frog, Ahrefs / Semrush, Sitebulb
Prioritize by impact — fix critical issues before optimizing nice-to-haves
Write for humans first — keyword-stuffed content hurts rankings
Check actual SERPs — search for target keywords to understand what Google currently rewards
Focus on search intent — match content type to what users actually want
Monitor competitors — see what top-ranking pages do well and identify gaps
Always curl the URL, read the body — HTTP 200 status means nothing if the SPA is returning HTML for /robots.txtor/sitemap.xml. Confirm content type and first few lines
Find the shared shell function — in Express/Node SSR apps, a shared HTML shell function is a force multiplier. One change fixes all pages
Dynamic sitemaps > static sitemaps — for apps with content arrays (guides, terms, products), generate the sitemap dynamically from those arrays so it stays current automatically
Cannot access Google Search Console or Analytics data
Cannot measure actual page speed (use Google Lighthouse separately)
Cannot check backlink profiles (recommend Ahrefs, Semrush, or Moz)
Cannot run full site crawls (recommend Screaming Frog or Sitebulb)
Cannot guarantee ranking improvements — SEO involves many factors
Cannot access pages behind authentication or paywalls