- Home
- Skills
- Marketing & SEO
- AI SEO, GEO & Agentic Commerce Architect
AI SEO, GEO & Agentic Commerce Architect
Modular 2026 workflow for Technical SEO, Schema.org JSON-LD (ProductGroup), AI Search Optimization (llmstxt.org v2 & .md Shadow Corpus), Streamable HTTP MCP servers, and Agentic Commerce protocols (Google/Shopify UCP & OpenAI ACP).
$12
Works with the AI Agents you already use
AI SEO, GEO & Agentic Commerce Architect
Example session with this skill installed
Use ai-seo-skill to audit our Next.js e-commerce product page and implement 2026-compliant multi-variant JSON-LD (ProductGroup with strikethrough sale pricing), llmstxt.org v2 discovery headers, and a /llms.txt endpoint.
- Read your context and instructions
- Compiled the ai seo geo
Summary
Audited the storefront product route and upgraded the SEO & AI discovery stack to the
2026 Google Search Central and
llmstxt.org v2 standards (DOMAIN_PROFILE = COMMERCE, layers: CORE_SEO + AI_DISCOVERY).
| Audit Area | Before | After (ai-seo-skill) |
|---|---|---|
| Multi-Variant JSON-LD | Deprecated AggregateOffer (disqualifies Merchant Listings) | @type: "ProductGroup" with variesBy, hasVariant, brand, and StrikethroughPrice |
| Description Sanitization | Raw regex tag strip (<p>A</p><p>B</p> → "AB", \u0026amp; bug) | 5-step normalizeDescription + XSS-safe encodeJsonLd (\u003C, \u003E, \u0026, \u0027) |
| Canonical URL | Included ?sort=asc&utm_source=... (Spider Trap risk) | Stripped all filter/tracking params; retained clean canonical path |
| AI Agent Discovery | None | Added /llms.txt, .md Shadow Corpus twin, and Link: rel="describedby" HTTP header |
1. Core Sanitization & XSS-Safe JSON-LD Serializer (src/lib/seo-utils.ts)
import { normalizeDescription, encodeJsonLd } from './core_utils';
export function buildProductGroupJsonLd(product: ProductDto) {
const payload = {
'@context': 'https://schema.org/',
'@type': 'ProductGroup',
name: product.title,
description: normalizeDescription(product.rawHtmlDescription, 300),
url: `https://example.com/products/${product.slug}`,
productGroupID: String(product.id),
brand: { '@type': 'Brand', name: product.brandName },
variesBy: ['https://schema.org/color', 'https://schema.org/size'],
hasVariant: product.variants.map((v) => ({
'@type': 'Product',
sku: v.sku,
name: `${product.title} - ${v.color} / ${v.size}`,
color: v.color,
size: v.size,
image: v.imageUrl,
offers: {
'@type': 'Offer',
url: `https://example.com/products/${product.slug}?variant=${v.id}`,
priceCurrency: v.currency,
price: v.priceMajor,
validFrom: v.saleValidFrom,
validThrough: v.saleValidThrough,
availability: v.inStock
? 'https://schema.org/InStock'
: 'https://schema.org/OutOfStock',
itemCondition: 'https://schema.org/NewCondition',
...(v.listPriceMajor > v.priceMajor
? {
priceSpecification: {
'@type': 'UnitPriceSpecification',
priceType: 'https://schema.org/StrikethroughPrice',
price: v.listPriceMajor,
priceCurrency: v.currency,
},
}
: {}),
},
})),
};
return encodeJsonLd(payload);
}
2. llmstxt.org v2 & Shadow Corpus Discovery Headers (middleware.ts)
res.headers.append(
'Link',
'<https://example.com/llms.txt>; rel="describedby"; type="text/markdown"'
);
res.headers.append(
'Link',
`<https://example.com${pathname}.md>; rel="alternate"; type="text/markdown"`
);
Next steps
- Run
npm testto execute the unit test suite verifyingnormalizeDescription,encodeJsonLd, andtoMinorUnits. - Optional: Enable Step 5 (
AGENTIC_COMMERCE) if you want to expose/.well-known/ucpand OpenAI ACPproducts.jsonlfeeds.
Connects securely to your tools. The creator never sees your data.
About this skill
Equip your AI Coding Agent to audit and implement 2026-standard Technical SEO, Schema.org JSON-LD, GEO / AI Search Optimization (llmstxt.org v2), Streamable HTTP MCP Servers, and Agentic Commerce (UCP & OpenAI ACP) across any web stack.
Why Developers Need This Skill
Most LLM coding assistants rely on outdated training data when generating structured data, sitemaps, or AI endpoints—resulting in deprecated schemas, XSS vulnerabilities, and broken agent feeds. ai-seo-skill enforces the latest 2026 production specifications:
- 2026 Google Search Central JSON-LD: Replaces deprecated
AggregateOfferwith@type: "ProductGroup"(variesBy+hasVariant), includesbrand, and placesvalidFrom/validThroughdirectly onOfferalongsideStrikethroughPrice. - Anti-XSS & Anti-Double-Encoding Utilities: Includes drop-in Python & TypeScript reference utilities for
normalizeDescription(5-step HTML block spacing + UTF-8 entity decode so&never becomes\u0026amp;),encodeJsonLd(hex-escaping<,>,&,'), andtoMinorUnits(strictly separatingprice_majorfor Schema.org from ISO 4217price_minorintegers for UCP/ACP). llmstxt.orgv2 & Markdown Shadow Corpus: Generates/llms.txt,/llms-full.txt, HTTPLink: <.../llms.txt>; rel="describedby"auto-discovery headers, per-page.mdShadow Corpus routes (rel="alternate"; type="text/markdown"reducing LLM token ingestion cost by ~95%), and granularrobots.txtrules separating AI Search Bots (OAI-SearchBot,ChatGPT-User,PerplexityBot) from AI Training Crawlers (GPTBot,CCBot).- Streamable HTTP MCP Server (
2026-07-28): ConfiguresPOST /mcpwith mandatory DNS Rebinding protection (Originvalidation rejectingOrigin: nullwith HTTP403,127.0.0.1local binding),Acceptnegotiation,_meta&Mcp-Session-Idpropagation, and OAuth 2.1 / Bearer auth. - Agentic Commerce (
UCP&OpenAI ACP): ImplementsGET /.well-known/ucpwith RFC 9421 EC P-256 JWKsigning_keys,/catalog/search,/lookup,/product, plus OpenAI/.well-known/acp.jsonand JSON Schema-validatedproducts.jsonlfeeds.
Domain-Adaptive & Modular (Progressive Disclosure)
The skill automatically detects your project's domain (COMMERCE, RESERVATION, SAAS_PLATFORM, CONTENT_PUBLISHING, or GENERIC_WEB) and loads only the reference layers you request—keeping context window usage minimal.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Marketing & SEO.
Single-file XML sitemaps capped at 10k URLs. Requires manual OAuth configuration for private MCP tools. Does not support non-Python/TS templates.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 11 days ago
- Software
Node.js
Git
Permissions
File Scopes
Reads existing routes/models to infer the project domain profile, writes SEO/JSON-LD/llms.txt/MCP/UCP endpoints, and runs the project's native test runner to verify compliance.