seo-tech-stack
Skill0 starsChanged 22 days agoMITView on GitHub
Detect site stack (Shopify, WordPress/Elementor/Woo, Next.js, Framer/Webflow) from the live page; run stack-specific tech-SEO checks + fixes. Use for "what is this site built on", "tech stack", "Shopify SEO issues", "WordPress technical SEO", "Next.js SEO". Generic audit → seo-technical/audit-site.
In this repository33 other files it gives its agents.
- AGENTS.mdAGENTS.md
- skills/setup-project-context/templates/CLAUDE.mdCLAUDE.md
- backend-architectureSkill
- build-review-loopSkill
- capability-intakeSkill
- competitor-market-analysisSkill
- context-auditSkill
- data-analysis-insightsSkill
- data-privacy-securitySkill
- dev-agentsSkill
- dev-patternsSkill
- frontend-system-designSkill
- idea-validatorSkill
- intellectual-property-licensingSkill
- international-legal-researchSkill
- legal-document-drafterSkill
- llm-councilSkill
- make-slidesSkill
- password-authentication-securitySkill
- privacy-data-protection-lawSkill
- project-estimation-planningSkill
- project-startSkill
- rea-researchSkill
- requirements-business-analystSkill
- secrets-credential-managerSkill
- setup-project-contextSkill
- shopify-ui-auditSkill
- site-incidentSkill
- sop-process-builderSkill
- task-dispatchSkill
- technical-feasibilitySkill
- tech-research-intelligenceSkill
- update-mdSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 16
Evidence only: every finding = URL + what was observed (header / DOM node / request / status / file:line).
- Very specific: line 17
Read-only: never edit live theme/site/settings, purge caches, toggle apps or push.
- Distinctive: line 97
No deploy without approval.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
seo-tech-stack: platform layer of technical SEO
Target: $ARGUMENTS
Detect stack from LIVE page → run that platform's checklist → findings with exact file / setting / app fix. Sample templates, don't crawl.
Generic 9-category audit = seo-technical. Full crawl + client deck = audit-site. This skill covers the platform quirks those two miss.
Hard rules
- Step 0 always: fetch https://developers.google.com/search/updates and diff against the baseline below. Report line:
Google Search updates checked (DD/MM/YYYY). - Evidence only: every finding = URL + what was observed (header / DOM node / request / status / file:line). No "probably". Unknown → say unknown.
- Read-only: never edit live theme/site/settings, purge caches, toggle apps or push. Pulling a theme or reading a repo locally is fine. Fixes = proposals routed through the project's own workflow (§5).
- No
org/the-orgin suggested client code: new snippet / file / ID / CSS class / comment names stay neutral (snippets/seo-head.liquid). - Known client domain → read the project's
CLAUDE.md+ memory note first (live theme ID, cache layers, workflow, past gotchas). Don't re-derive. - Tracking/pixel inventory: the chrome-devtools profile ad-blocks clarity / consent / otlp. Use clean Playwright Chromium with a real UA for that part.
Baseline Google facts (written 21/10/2026, re-verify in step 0)
- Googlebot indexes the first 2MB of each HTML file. CSS/JS are fetched separately, same cap.
- JS: non-200 pages may skip rendering. A
noindexin the original HTML may skip rendering, so JS can't remove it. Canonical in raw HTML must equal rendered. rel=canonicalis honoured only in<head>. Robots meta is honoured in<body>too.- Faceted nav: for low-value filters prefer robots.txt disallow or fragments; canonical is weak long-term. Empty filter combos → 404.
- FAQ rich result is gone (May 2026). Search doesn't use
llms.txt(Jun 2026). Breadcrumb rich result is desktop-only. New "back button hijacking" spam policy (Apr 2026). - Merchant: org-level shipping + return policy markup is supported.
Product.categoryand sale-price validity dates were added Jul 2026.
1 · Detect
Sample: home, PLP/category, PDP/post, static page, 1 filtered/sorted URL, /robots.txt, sitemap, 404 probe /zz-404-<rand>.
chrome-devtools MCP (default): navigate_page → list_network_requests + get_network_request (doc headers + raw body) → evaluate_script:
() => { const q=s=>[...document.querySelectorAll(s)];
return { url:location.href, shopify:window.Shopify?{shop:Shopify.shop,theme:Shopify.theme,root:Shopify.routes?.root,cur:Shopify.currency?.active}:null,
next:window.next?.version||(window.__NEXT_DATA__?'pages':(self.__next_f?'app':null)),
wp:!!q('link[href*="/wp-content/"],script[src*="/wp-content/"]').length, elementor:!!q('[data-elementor-type]').length,
gen:q('meta[name=generator]').map(m=>m.content), wf:document.documentElement.dataset.wfSite||null,
canon:q('link[rel=canonical]').map(l=>l.href+(l.closest('head')?'':' [BODY]')),
robots:q('meta[name=robots],meta[name=googlebot]').map(m=>m.content),
ld:q('script[type="application/ld+json"]').flatMap(s=>{try{const j=JSON.parse(s.textContent);return [].concat(j['@graph']||j).map(x=>x['@type'])}catch{return['INVALID']}}),
h1:q('h1').length, links:q('a[href]').length, nodes:q('*').length, bytes:document.documentElement.outerHTML.length } }Raw HTML: the doc response body, or curl -sL -A '<Googlebot smartphone UA>'. Cloudflare may block curl, so fall back to the browser.
| Layer | Signals |
| Shopify | window.Shopify, cdn.shopify.com, /cdn/shop/, x-shopify-stage, powered-by: Shopify |
| WordPress | /wp-content/, Link: <…/wp-json/>; rel="https://api.w.org/", generator, Yoast / Rank Math HTML comments |
| Next / React | __NEXT_DATA__ (Pages), self.__next_f (App), /_next/, x-powered-by: Next.js, x-nextjs-cache; empty #root = SPA |
| Builders | Framer generator / framerusercontent.com; Webflow data-wf-site; Wix x-wix-request-id; Squarespace static1.squarespace.com |
| Cache / CDN | cf-ray, cf-cache-status, x-litespeed-cache, x-cache-status, x-varnish, x-vercel-cache, age |
| Host | x-vercel-id, platform: hostinger, server: nginx/LiteSpeed, Cloudways Varnish, x-nf-request-id |
Record: theme + version, builder, SEO owner (who emits title / canonical / robots / JSON-LD, and duplicate count), cache layers, CDN/WAF, host, render mode (raw vs rendered parity), third parties (checkout, reviews, currency/geo, pixels, chat, deferral engines).
2 · Stack card
Platform: <name + version> · Theme/app: <name, id, version> · Builder: <…> · SEO owner: <plugin/app/theme/snippet>
Render: SSR|SSG|ISR|CSR|hybrid (raw-vs-rendered: title/canon/robots/H1/copy/links/JSON-LD = same|diff)
Cache/CDN: <layers, with the header seen> · Host: <…> · Third-party: checkout/reviews/currency/pixels/chat/deferral
Sampled: <URLs>3 · Platform checklist
Load ONLY the matching ref(s): Shopify → references/shopify.md · WordPress (+Elementor/Woo) → references/wordpress.md · Next.js / React SPA → references/nextjs.md · Framer / Webflow / Wix / static / migrations → references/other.md.
Hybrid stacks (headless Shopify + Next, WP blog in a subfolder, a migration mid-flight) → load both.
4 · Cross-stack layer (always, cheap)
- Host variants: http/https × www/apex → one 301 hop to one host, with the same stack on every variant. (Seen: www still Framer while apex was WP.)
- Platform/staging hosts (
*.myshopify.com,*.vercel.app,*.hostingersite.com, nip.io/sslip.io,*.webflow.io,*.framer.app): each should 301 or be noindexed. Prod must not carry staging noindex (check both meta andX-Robots-Tag). - CDN/WAF: Googlebot UA gets 200, not a challenge. Check for a Cloudflare managed robots.txt block or AI-bot blocking at zone level (one zone can front many client hosts).
- Cache truth:
age/x-*cache*headers decide whether the evidence is fresh. A fix may be live but still cached; note the bypass used. - Weight: raw doc bytes vs the 2MB cap, DOM nodes, duplicated menus/sections.
- Parity: final URL == requested (watch for a client-side bounce to
/login). Title, canonical, robots, H1, copy,<a href>links and JSON-LD should be the same raw and rendered. - Deferral engines (
d-src,type="litespeed/javascript",type="load", interaction-gated loaders): Lighthouse understates the cost, so use CrUX field data. Check whether pixels fire with no interaction. - Back-button hijack / forced-redirect scripts (exit-offer apps) → spam-policy risk.
- Cloaking: diff Googlebot UA vs browser. IP cloaking can't be caught by UA; GSC URL Inspection is ground truth (
gsc_inspect_url/hub_gsc_inspectif connected).
5 · Report
## Stack card (§2)
## Findings
| # | Impact | Finding | Evidence (URL → observed) | Stack fix (file / setting / app) | Route |
## Hand-offs
Google Search updates checked (DD/MM/YYYY)- Impact:
- P0: deindex or site-wide (noindex, robots block, canonical → home, auth bounce, every route soft-404).
- P1: template-wide duplication or crawl waste, broken sitemap, failing field CWV.
- P2: one template or app.
- P3: hygiene.
- Route (the owner executes, never this skill):
- Shopify: pull LIVE by ID → edit locally → push to a non-live theme → client publishes.
- WP: staging → PR → client cutover.
- Next/React: PR-only (a merge to an auto-deploying main is a deploy).
- No deploy without approval.
- Hand-offs:
- Generic crawl/index/CWV/security headers →
seo-technical - JSON-LD validity →
seo-schema - hreflang matrix →
seo-hreflang - Sitemap gaps →
seo-sitemap - Product/merchant →
seo-ecommerce - Full crawl + deck →
audit-site - Hack, cloaking, spam pages indexed →
site-incident; framework CVE / code review →security-audit
- Generic crawl/index/CWV/security headers →
- Output: chat by default. When called from
audit-site, write<output_dir>/02b-tech-stack.md.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

