Complete SEO + UX + Conversion audit agent. Crawls a website, scores every dimension 0-100, produces prioritized action plan. Covers technical SEO, on-page, content quality, E-E-A-T, user personas, buying intent, navigation, Core Web Vitals, schema, AI search readiness (AEO/GEO), AI writing detection, and conversion paths. Use when user says "SEO audit", "audit this website", "site audit", "check SEO", "website review", "why am I not ranking", "my traffic dropped", "not showing up in Google", "page speed", "core web vitals", "crawl errors", "indexing issues", "my SEO is bad", "help with SEO", "content audit", or provides a URL for analysis. Works for SaaS, e-commerce, blogs, local business, apps, marketplaces. See programmatic-seo for pages at scale, schema-markup for structured data, ai-seo for AI search details, page-cro for page-level CRO.
Not a surface scan. A full teardown of every URL, every tag, every user path, every conversion blocker, every missed ranking opportunity.
Combines technical SEO crawling, UX heuristic analysis, content scoring, conversion funnel mapping, competitor positioning, AI search readiness, and AI writing detection into one comprehensive audit.
Check for product marketing context first:
If .agents/product-marketing-context.md exists (or .claude/product-marketing-context.md in older setups), read it before asking questions. Use that context and only ask for information not already covered.
Before auditing, understand:
If user just drops a URL, assume full audit, auto-detect pages (cap 50), all dimensions.
Schema Detection: web_fetch and curl strip <script> tags. JSON-LD injected by CMS plugins (Yoast, RankMath, AIOSEO) won't appear. Use browser tool, Rich Results Test, Firecrawl, or Screaming Frog instead. Always caveat schema findings when using non-rendering tools.
Core Web Vitals: PageSpeed Insights API or Chrome UX Report for real field data. Lab data is directional, not definitive.
Firecrawl preferred (renders JS). Fallback: web_fetch (rate-limited, caveat schema). For each URL extract: status code, response time, title, meta description, meta robots, canonical, headings (H1-H6), images (src + alt), internal links (href + anchor), external links, schema/JSON-LD, OG tags, hreflang, word count, language.
Crawlability, indexation (compare indexed vs expected), Core Web Vitals, mobile responsiveness, security (HTTPS + headers), URL structure, redirect chains, broken links.
Score per page (0-10 per element, weighted to 0-100): title tags, meta descriptions, heading structure, image optimization, internal linking, keyword targeting, cannibalization check.
Flag: excessive em dashes, AI phrases ("delve into", "it's important to note"), formulaic structure, uniform sentence length, lack of personal voice, generic filler paragraphs. Google's December 2025 update penalizes these patterns.
E-E-A-T (0-100), content depth vs SERP competition, readability, freshness, duplicate/thin detection, topical authority assessment.
Persona inference from site language/design, navigation (7±2, breadcrumbs, 3-click rule), visual hierarchy (5-second test), cognitive load (Hick's Law), mobile UX, accessibility, trust signals.
Page-by-page intent classification, funnel walkthrough simulation, CTA scoring (visibility + copy + intent match + singularity), friction identification, pricing page analysis.
Robots.txt AI bot access (GPTBot, ClaudeBot, PerplexityBot, OAI-SearchBot, Bytespider), llms.txt presence and quality, BLUF formatting, structured extractable answers, content citability, entity SEO (sameAs links, Knowledge Panel, Wikidata), brand mention presence across platforms (Reddit, YouTube, Wikipedia, LinkedIn), Featured Snippet and PAA targeting, IndexNow implementation, social meta tags.
For top 3 competitors: schema comparison, content gaps, depth comparison, speed comparison, backlink-worthy assets.
If DeepSeek/Gemini/OpenAI MCPs connected: send content for cross-model quality scoring, persona alignment, competitor comparison.
SaaS: Thin product pages, blog disconnected from product, missing comparison pages, no glossary content.
E-commerce: Thin categories, duplicate product descriptions, missing product schema, faceted navigation duplicates, out-of-stock mishandling.
Content/Blog: Outdated content, keyword cannibalization, no topical clusters, poor interlinking, missing author pages.
Local Business: Inconsistent NAP, missing LocalBusiness schema, no GBP optimization, no location pages.
App Landing Pages: No SEO depth beyond one page, missing SoftwareApplication schema, no content strategy, store listing not optimized.
Marketplace: UGC quality issues, thin listings, duplicates, missing ItemList/Product schema.
| Score | Grade | Meaning |
|---|---|---|
| 90-100 | A+ | Industry-leading |
| 80-89 | A | Strong, optimize |
| 70-79 | B | Good, notable gaps |
| 60-69 | C | Significant gaps |
| 50-59 | D | Major issues |
| 0-49 | F | Critical overhaul |
Overall = weighted average: Technical 25%, On-Page 20%, Content 20%, UX 15%, Conversion 10%, AI Readiness 10%.
End every report with:
If you could only fix 3 things this week, which 3 would move the needle most?
| File | Purpose |
|---|---|
| technical-crawl-protocol.md | Crawling with Firecrawl, web_fetch, Playwright |
| technical-audit-checklist.md | Crawlability, indexation, speed, mobile, security |
| on-page-scoring.md | Title, meta, heading, image, link scoring rubrics |
| content-scoring.md | E-E-A-T, depth, readability, freshness |
| ai-writing-detection.md | AI content patterns, detection, remediation |
| ux-analysis-framework.md | Nielsen heuristics, persona, navigation, trust |
| conversion-analysis.md | Intent mapping, funnel, CTA, friction |
| ai-search-readiness.md | Bot access, BLUF, citability, platforms |
| report-template.md | Executive summary, detailed report templates |