Shopify Admin Agentic Description Enrichment

Purpose

When an AI agent decides whether to recommend a product, it quotes the description to justify the match. One-line or empty descriptions give it nothing — so it can't confirm the product fits the shopper's intent and moves on. This skill finds products with thin descriptions (below a character threshold or missing key attributes) and rewrites them into structured copy: a benefit-led opening line, then concrete facts (material, fit/sizing, dimensions, use-cases, care) drawn from the product's own metafields/options/type. Fixes listing-description-quality and strengthens agentic-listing-schema.

Prerequisites

• Authenticated Shopify CLI session (shopify auth login --store <domain>)
• Required API scopes: read_products, write_products

Parameters

All skills accept these universal parameters:

Skill-specific parameters:

Safety

⚠️ Step 3 (productUpdate) replaces descriptionHtml on live products. This is content the merchant may have hand-written. Always run dry_run: true, export the before/after, and have a human approve the rewrites before committing. Default only touches products under min_chars.

Workflow Steps

1. OPERATION: products — query Inputs: first: 100, optional filter; fields descriptionHtml, title, productType, options, tags, metafields(first: 30); paginate. Expected output: Products whose stripped-text description is below min_chars, plus the structured facts available to enrich from.

2. COMPUTE (no API): for each thin product, draft descriptionHtml: a one-line benefit hook + a short facts list built ONLY from real product data (no invented specs), honoring brand_voice. Emit a before/after preview.

3. OPERATION: productUpdate — mutation Inputs: { id, descriptionHtml } per approved product. Expected output: Updated product; collect userErrors.

GraphQL Operations

# products:query — validated against api_version 2025-01
query EnrichProducts($first: Int!, $after: String, $query: String) {
  products(first: $first, after: $after, query: $query) {
    edges {
      node {
        id
        title
        descriptionHtml
        productType
        tags
        options { name values }
        metafields(first: 30) { edges { node { namespace key value type } } }
      }
    }
    pageInfo { hasNextPage endCursor }
  }
}

# productUpdate:mutation — validated against api_version 2025-01
mutation EnrichProductDescription($input: ProductInput!) {
  productUpdate(input: $input) {
    product { id descriptionHtml }
    userErrors { field message }
  }
}

Session Tracking

Claude MUST emit the following output at each stage. This is mandatory.

On start, emit:

╔══════════════════════════════════════════════╗
║  SKILL: <skill name>                         ║
║  Store: <store domain>                       ║
║  Started: <YYYY-MM-DD HH:MM UTC>             ║
╚══════════════════════════════════════════════╝

After each step, emit:

[N/TOTAL] <QUERY|MUTATION>  <OperationName>
          → Params: <brief summary of key inputs>
          → Result: <count or outcome>

If dry_run: true, prefix every mutation step with [DRY RUN] and do not execute it.

On completion, emit:

For format: human (default):

══════════════════════════════════════════════
OUTCOME SUMMARY
  <Metric label>:   <value>
  Errors:           0
  Output:           <filename or "none">
══════════════════════════════════════════════

For format: json, emit:

{
  "skill": "<skill-slug>",
  "store": "<domain>",
  "started_at": "<ISO8601>",
  "completed_at": "<ISO8601>",
  "dry_run": false,
  "steps": [
    {
      "step": 1,
      "operation": "<OperationName>",
      "type": "query",
      "params_summary": "<string>",
      "result_summary": "<string>",
      "skipped": false
    }
  ],
  "outcome": {
    "metric_key": 0,
    "errors": 0,
    "output_file": null
  }
}

Output Format

human: count enriched + a before/after CSV (product, old_len, new_len, new_description). json: { products_enriched, products_skipped, errors, output_file }.

Error Handling

Best Practices

• Enrich from facts the store already holds (metafields, options, type) — never fabricate materials, certifications, or measurements.
• Front-load the attributes shoppers search by; agents weight the first sentence heavily when matching intent.
• This pairs with shopify-admin-agentic-metafields-setup: structured metafields make the descriptions both richer and machine-filterable.
• Always have a human approve the dry-run diff for hero/bestseller products before committing.