Shopify Admin Vendor Consolidation

Purpose

Walks every product in the catalog, normalizes the vendor field, and clusters near-duplicates such as Acme, ACME, Acme Inc, and Acme (trailing whitespace). Surfaces a recommended canonical form per cluster and the count of products that would migrate. Vendor sprawl breaks vendor-based reports, navigation, and supplier reconciliation. Read-only — no mutations; output is the worklist for a follow-up consolidation.

Prerequisites

• Authenticated Shopify CLI session: shopify store auth --store <domain> --scopes read_products
• API scopes: read_products

Parameters

Safety

ℹ️ Read-only skill — no mutations are executed. Safe to run at any time. The skill produces a recommendation worklist; consolidation must be applied through a separate, reviewed workflow.

Workflow Steps

1. OPERATION: products — query Inputs: first: 250, query: <built from status_filter>, select vendor, id, title, status, pagination cursor Expected output: Every product with its vendor string; paginate until hasNextPage: false

2. Build a frequency map of distinct vendor strings → product count. Normalize each vendor with: trim whitespace, collapse multiple spaces, strip configured suffixes, lowercase for comparison.

3. Cluster vendor strings whose normalized form has a Levenshtein ratio above similarity_threshold. Pick canonical per cluster as the most-used variant; tie-break on shortest, then alphabetical.

4. Filter clusters with fewer than min_cluster_size distinct strings. Compute migration impact: number of products that would move to the canonical form.

5. Sort clusters by migration impact descending so the highest-leverage merges surface first.

GraphQL Operations

graphqlCopy code
# products:query — validated against api_version 2025-01
query AllVendors($query: String, $after: String) {
  products(first: 250, after: $after, query: $query) {
    edges {
      node {
        id
        title
        vendor
        status
        productType
        updatedAt
      }
    }
    pageInfo {
      hasNextPage
      endCursor
    }
  }
}

Session Tracking

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

On start, emit:

Copy code
╔══════════════════════════════════════════════╗
║  SKILL: Vendor Consolidation                 ║
║  Store: <store domain>                       ║
║  Started: <YYYY-MM-DD HH:MM UTC>             ║
╚══════════════════════════════════════════════╝

After each step, emit:

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

On completion, emit:

For format: human (default):

Copy code
══════════════════════════════════════════════
VENDOR CONSOLIDATION REPORT
  Products scanned:        <n>
  Distinct vendors:        <n>
  Clusters detected:       <n>
  Products to migrate:     <n>

  Top clusters by impact:
    Canonical: "<name>"  Variants: <n>  Products: <n>
      "<variant 1>"  (<count>)
      "<variant 2>"  (<count>)
  Output: vendor_consolidation_<date>.csv
══════════════════════════════════════════════

For format: json, emit:

jsonCopy code
{
  "skill": "vendor-consolidation",
  "store": "<domain>",
  "products_scanned": 0,
  "distinct_vendors": 0,
  "clusters": [
    {
      "canonical": "Acme",
      "variants": [
        { "value": "Acme", "products": 0 },
        { "value": "ACME", "products": 0 },
        { "value": "Acme Inc", "products": 0 }
      ],
      "products_to_migrate": 0
    }
  ],
  "output_file": "vendor_consolidation_<date>.csv"
}

Output Format

CSV file vendor_consolidation_<YYYY-MM-DD>.csv with columns: cluster_id, canonical_vendor, variant_vendor, is_canonical, product_count, sample_product_id, sample_product_title, status

Error Handling

Best Practices

• Start with the default similarity_threshold: 0.88. Below 0.85, false positives multiply quickly.
• Manually review every cluster before applying — ABC Corp and ABC Co. may or may not be the same supplier in your books.
• Use ignore_suffixes to absorb legal-entity noise (Inc, LLC) which rarely changes the actual vendor identity.
• After review, drive consolidation via a separate update workflow (for example, an internal product-update script) and re-run this audit until clusters drop below min_cluster_size.
• Pair with product-data-completeness-score to track vendor-field cleanliness over time alongside other catalog quality metrics.