A Claude Code skill that reads your theme files and your rendered HTML, runs 80+ checks across performance, accessibility, conversion, third-party app overhead, classic search, and AI-search visibility. Free v2.1 is on GitHub. v3 Premium is for Plus merchants and agencies who need the three composite pillars.
Score
87/100
Grade
B
Built for themes on
v2.1 is the open-source baseline. v3 Premium adds the three pillars Plus merchants and agency teams need to justify the spend.
Reply within 24 hours.
Already using v2.1? Upgrades to v3 keep your existing skill files — premium adds two new markdown files.
Why this exists now
A theme audit in 2026 cannot stop at Lighthouse and meta tags. Buyers research products in three different surfaces, each with its own retrieval logic. This skill audits all three.
Google · Bing · DuckDuckGo
Classic crawl & index search. Still the largest traffic source for most stores. Wins or loses on titles, meta, canonicals, schema, hreflang, Core Web Vitals.
ChatGPT · Claude · Perplexity · Gemini
Answer Engine Optimization. The user never sees a results page. If your product page is not quotable, you are invisible at the moment of decision.
PropertyValueAI Overviews · Copilot · LLM shopping
Generative Engine Optimization. AI-powered surfaces either cite your store, or it doesn't exist to the buyer. Citation depends on schema, crawlability for AI bots, and trust signals.
sameAsLayer 0 Before any of the three search layers can rank you, your store has to load. App overhead is where most stores lose this round.
See app checks →What it checks
Render-blocking scripts, lazy loading, image dimensions, LCP preload, oversized CSS, N+1 queries inside collection loops.
Skip-to-content link, focus trap on cart drawer, ARIA labels on icon buttons, decorative vs informational alt text, keyboard reachability.
Most Shopify stores die from app bloat, not theme bloat. Detects 20+ apps in compiled HTML/JS and scores each by load strategy, bundle weight, and above-fold impact.
APP-C · APP-H · APP-M · APP-L
Zero-results state, password page brand, trust elements in footer, breadcrumbs.
M-class · −2 each
Title, meta, canonical, sitemap, multiple H1s, Open Graph, Twitter Cards, hreflang.
S-C · S-H · S-M
FAQPage, HowTo, Speakable, Article + author, factual openings, PropertyValue specs.
AEO-C · AEO-H · AEO-M
Robots.txt for AI crawlers, llms.txt overview, Organization schema with social profiles, BreadcrumbList, shipping & returns in Product schema, image URLs & alt for multimodal models.
GEO-C · GEO-H · GEO-M · GEO-L
Every JSON-LD schema your store should emit, confirmed or flagged.
Sample output
v2.1 returns a single 0–100 score with split-mode option. v3 Premium returns a three-sub-score composite. Both shown below, lifted from real audits.
bloom-theme · audit.md · v2.1
v2.1 · FreeTheme: Bloom (Dawn-based, Online Store 2.0)Mode: Full auditTechnical & CRO Score: 69/100, Grade: CSEO / AEO / GEO Score: 58/100, Grade: D ━━ Critical · Technical (1) ━━C6: Images missing height attribute (CLS) sections/header.liquid:192,237 sections/bloom-hero.liquid:18,44 ━━ Critical · SEO / AEO / GEO (3) ━━S-C2: Meta description missing on collection pages layout/theme.liquid:34AEO-C1: FAQ accordion present, no FAQPage schema sections/faq.liquid:1-80GEO-C1: robots.txt blocks GPTBot, ClaudeBot, PerplexityBot templates/robots.txt.liquid:12-18 ━━ High Priority (5) ━━H3: No preload hint for LCP hero imageH4: base.css at 81KB loaded on every pageS-H1: Header logo wrapped in <h1>AEO-H1: Product specs buried in proseGEO-H1: Organization schema missing sameAs ━━ Third-Party App Overhead (4) ━━APP-C1: Klaviyo onsite.js loads sync in <head> (+420ms FCP) rendered-html:head:38APP-C2: Gorgias chat above-fold on mobile (+1.2s LCP) snippets/gorgias-chat.liquid:4APP-H1: Judge.me widget container has no reserved height (CLS) sections/main-product.liquid:218APP-M1: Hotjar recording 100% of sessions in production layout/theme.liquid:91 ━━ What This Theme Does Well (9) ━━ ✓ All JS uses defer, zero render-blocking ✓ Skip-to-content link present ✓ Online Store 2.0 throughout ✓ Product structured_data filter used ✓ Cart drawer trapFocus implemented ✓ Hreflang for all five Markets locales ✓ Canonical handles ?variant= correctly ✓ dateModified on article schema ✓ CSS custom properties for brand colors ━━ Estimated Impact After Fixes ━━ Lighthouse Performance: +8 to +15 LCP: -400 to -800ms CLS: substantial reduction AEO citations: FAQ + factual summaries are highest leverage GEO visibility: AI crawler access + Org schema unlock brand disambiguation
bloom-theme · audit-v3.md · composite
v3 PREMIUMTheme: Bloom (Dawn-based, Online Store 2.0, Plus)Mode: v3 Premium · Composite audit ━━ Composite Sub-Scores ━━ Theme Health 78/100 (perf · a11y · CRO · apps · SEO/AEO/GEO) Plus Architecture 54/100 (checkout ext · Functions · Markets · B2B) Agent Readiness 31/100 (JSON-LD · policies · cart · llms.txt) Business Economics 61/100 *partial (CVR & AOV pasted, HAR pending) ──────────────────────────────────────────── COMPOSITE 57/100 → Grade F ━━ Critical · Plus Architecture (2) ━━PA-C1: No checkout UI extensions; using deprecated checkout.liquid checkout.liquid:1PA-C2: Markets enabled but currency selector missing on PDP sections/main-product.liquid ━━ Critical · Agent Readiness (3) ━━AR-C1: Price & availability rendered via JS, not in JSON-LDAR-C2: Shipping/returns page is JS-modal, no server URLAR-C3: llms.txt missing ━━ Business Economics · ROI Roadmap (top 5) ━━ 1. Defer Klaviyo onsite.js → est. +0.3% CVR ≈ $4,200/mo 2. Gate Gorgias on intent → est. -0.9s LCP ≈ $3,800/mo 3. Add Product JSON-LD vars → AI agent reach unlocked 4. Reserve Judge.me height → CLS -0.06 → +0.1% CVR 5. Compress base.css → -54KB/page · -180ms FCP Every line above is computed from your pasted field data.Never fabricated. See methodology in business-economics.md.
By the numbers
Every check the skill runs comes from production theme work for 32+ brands. Three published example reports — C, F, and A grades — show the full range. No fluff, no abstract advice, every finding ships with a working fix.
80+
Total checks
Across performance, accessibility, CRO, apps, SEO, AEO, and GEO.
35+
SEO + AEO + GEO checks
A dedicated checklist for classic search, ChatGPT/Claude/Perplexity citation, and Google AI Overviews.
9
Markdown files
Eight drop into your theme root, plus a CHANGELOG. No build step. No subscription.
3
Audit modes
Full, Split-score (Technical/CRO vs SEO/AEO/GEO), and Quick-Wins ranked by impact ÷ effort.
Performance
13 checks
Accessibility
5 checks
Apps
21 checks
CRO
4 checks
SEO
18 checks
AEO
10 checks
GEO
9 checks
Install
Clone or download from GitHub.
git clone https://github.com/
Eight markdown files at the theme root.
your-shopify-theme/ ├── CLAUDE.md ├── audit-checklist.md ├── seo-aeo-geo-checklist.md ├── apps-audit.md ├── before-after.md ├── liquid-patterns.md ├── deprecated-apis.md └── scoring.md
Open the project. Ask for an audit.
Audit this Shopify theme using the audit skill. Cover performance, app overhead, accessibility, conversion, SEO, AEO, and GEO.
Other prompts
Quick-Wins · pre-BFCM
Run the audit skill in quick-wins mode. I have one hour before BFCM.
App overhead only
Audit only third-party app overhead using apps-audit.md.
Search audit only
Run an SEO, AEO, and GEO audit. Use split-score mode.
Single area
Audit only the product page for AEO, FAQ + factual summaries + spec structure.
Live store via MCP
Connect via the Shopify MCP. Audit what you see. Note skipped source-level checks.
Use cases
Catch regressions before going live. Run every check, fix what matters, ship clean.
Produce a credible, sourced audit deliverable in minutes, not days.
Audit migrated themes against modern Shopify expectations. No deprecated APIs.
Find LCP, CLS, and render-blocking issues before peak traffic destroys conversion.
Confirm the store can be cited by ChatGPT, Claude, Perplexity, and AI Overviews.
Verify Product, FAQ, Breadcrumb, Organization, Article, HowTo schemas across the catalog.
Confirm the merchant's intent for GPTBot, ClaudeBot, PerplexityBot is reflected in robots.txt.
Point Claude at one section file for a tight, focused audit. Ship the fix the same hour.
Detects 20+ Shopify apps in your compiled HTML, ranks each by LCP cost, and tells you which ones to defer, gate, or uninstall.
Re-rank every finding by impact ÷ effort. Star-rated table at the top. A 5-minute High beats a half-day Critical, so you fix the right thing first.
The skill is free. The premium pillars are paid. Done-for-you audits are a fixed-scope project.
The audit skill itself. Open source, MIT licensed, yours to fork.
Three paid pillars on top of v2.1. Sub-score composite, not a single number.
Everything in Free, plus
Reply within 24 hours · USD invoice
/ per store
A fixed-scope engagement. You ship the merchant URL, I ship the report.
Quick answers
v2.1 is the free, open-source baseline. It runs 80+ checks across performance, accessibility, CRO, third-party app overhead, SEO, AEO, and GEO, and returns a single 0–100 score. v3 Premium adds three paid pillars on top of v2.1 — Plus Architecture (checkout extensibility, Functions, Markets, B2B, multi-storefront drift), Agentic Commerce Readiness (whether AI shopping agents can read your store), and Business Economics (live-data ROI mapping every finding to revenue). v3 produces a three-sub-score composite (Theme Health · Plus Architecture · Agent Readiness · Business Economics) with a composite grade, instead of one number.
v3 Premium is distributed by email only. Send a request to tanujrajput.dev@gmail.com with your store URL, Shopify plan, and what you're trying to fix. Reply within 24 hours with access and a USD invoice.
Yes. The skill is designed to run against local theme files. The Shopify MCP is optional, useful for live-store reads (shop info, products, analytics), but every source-level check works offline.
Audit in two passes. First layout/ + sections/, then snippets/ + templates/. Claude will add findings to the previous report instead of starting over.
Yes, and the report will note when it detects a Theme Store base. Any issues found are likely from merchant customizations, not the original theme author. Scoring stays the same.
For most stores, yes, a meaningful share of product-research traffic is now shifting to ChatGPT, Claude, Perplexity, and Google AI Overviews. The fixes themselves (FAQ schema, factual openings, Organization schema, AI-crawler access) are low-risk and frequently improve classic SEO too.
Yes. v2.1 is MIT licensed. Use it, modify it, sell audits with it. Attribution appreciated but not required. v3 Premium is a paid, per-merchant license — contact for terms.
Lighthouse measures symptoms. This skill reads your Liquid and tells you the cause, the exact file, line, and the fix. It also covers CRO, schema, accessibility, and AI-search visibility, which Lighthouse does not.
Yes. The apps-audit checklist greps the rendered HTML and JS for 20+ known app signatures — Klaviyo, Judge.me, Loox, Yotpo, Rebuy, Gorgias, Tidio, Smile.io, Recharge, Privy, Nosto, Hotjar, Microsoft Clarity, and more. Each detected app is scored by load strategy (sync vs deferred), bundle weight, and above-fold impact, then ranked alongside the theme findings so you see which apps are actually costing you LCP versus earning their weight.
Theme files alone get you ~90% of the audit. For the app overhead checks, the skill is most accurate when it can also read rendered HTML (paste a curl of your homepage and PDP, or connect via the Shopify MCP). Without that, it still flags hard-coded app snippets in your sections and snippets, plus theme app extensions in extensions/.
Quick-Wins mode re-ranks every finding by impact ÷ effort instead of raw severity, and leads the report with a star-rated table of the top 10 highest-ROI fixes before the standard Critical → High → Medium → Low breakdown. Use it when you are time-boxed, pre-BFCM, an agency on a fixed-scope sprint, or a few hours from a launch. Trigger it with phrases like "what should I fix first", "biggest bang for buck", or just "run in quick-wins mode".
Shopify developer and founder of EcomLifters, a Shopify development agency. 5+ years shipping custom themes and apps for 32+ brands across India and Canada. Built this skill from the audits I actually run, the bugs I keep finding, and the AI-search shifts I see hitting ecommerce right now.
Ship it
Free v2.1 is on GitHub. v3 Premium is one email away.