Metafield Definition Audit — Shopify Admin Skills

Read-only: enumerates every metafield definition across all owner types and flags unused, undocumented, or duplicate-key definitions.

Purpose

Inventories every metafield definition (PRODUCT, VARIANT, CUSTOMER, ORDER, COLLECTION, COMPANY, LOCATION, and others) and flags definitions that are unused (zero values stored), undocumented (missing description), or share a namespace.key collision across owner types. Definition sprawl is a leading source of theme/app bugs and slow Admin search. Read-only — no mutations. Provides the data foundation for a definition-cleanup workflow.

Prerequisites

Authenticated Shopify CLI session: shopify store auth --store --scopes read_products,read_customers,read_orders,read_inventory

API scopes: read scopes for any owner types in scope (defaults: read_products, read_customers, read_orders)

Parameters

Parameter	Type	Required	Default	Description
store	string	yes	—	Store domain (e.g., mystore.myshopify.com)
owner_types	string	no	all	Comma-separated owner types to scan (e.g. `PRODUCT,CUSTOMER`); `all` scans every supported type
flag_unused	bool	no	true	Flag definitions whose `metafieldsCount` is zero
flag_undocumented	bool	no	true	Flag definitions with empty/null `description`
flag_duplicates	bool	no	true	Flag `namespace.key` pairs that exist on more than one owner type
format	string	no	human	Output format: `human` or `json`

Safety

> ℹ️ Read-only skill — no mutations are executed. Safe to run at any time. No metafield definitions are deleted, updated, or pinned by this skill.

Workflow Steps

Determine the list of owner types to scan from owner_types (default: full list).

OPERATION: metafieldDefinitions — query

Inputs: For each owner type: first: 250, ownerType: , select id, namespace, key, name, description, type { name }, pinnedPosition, metafieldsCount, validations { name value }, pagination cursor

Expected output: All definitions per owner type with usage counts; paginate until hasNextPage: false

Build flag set per definition:

unused — metafieldsCount == 0 and flag_unused: true

undocumented — description is null or empty and flag_undocumented: true

duplicate_key — namespace.key appears on more than one owner type and flag_duplicates: true

Group results by owner type for the report and emit per-flag summaries.

GraphQL Operations

# metafieldDefinitions:query — validated against api_version 2025-01
query MetafieldDefinitionAudit($ownerType: MetafieldOwnerType!, $after: String) {
  metafieldDefinitions(first: 250, after: $after, ownerType: $ownerType) {
    edges {
      node {
        id
        namespace
        key
        name
        description
        ownerType
        pinnedPosition
        metafieldsCount
        type {
          name
          category
        }
        validations {
          name
          value
          type
        }
        access {
          admin
          storefront
        }
      }
    }
    pageInfo {
      hasNextPage
      endCursor
    }
  }
}

Session Tracking

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

On start, emit:

╔══════════════════════════════════════════════╗
║  SKILL: Metafield Definition Audit           ║
║  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>

On completion, emit:

For format: human (default):

══════════════════════════════════════════════
METAFIELD DEFINITION AUDIT
  Owner types scanned:    <n>
  Total definitions:      <n>

  By owner type:
    PRODUCT:    <n>   (unused: <n>, undocumented: <n>)
    VARIANT:    <n>   (unused: <n>, undocumented: <n>)
    CUSTOMER:   <n>   (unused: <n>, undocumented: <n>)
    ORDER:      <n>   (unused: <n>, undocumented: <n>)

  Flags:
    Unused definitions:        <n>
    Undocumented definitions:  <n>
    Duplicate keys:            <n>

  Examples:
    PRODUCT  custom.swatch_hex      unused (0 values)
    CUSTOMER custom.vip_tier        undocumented
    ORDER+CUSTOMER  custom.notes    duplicate key across owner types
  Output: metafield_def_audit_<date>.csv
══════════════════════════════════════════════

For format: json, emit:

{
  "skill": "metafield-definition-audit",
  "store": "<domain>",
  "owner_types_scanned": 0,
  "total_definitions": 0,
  "unused_definitions": 0,
  "undocumented_definitions": 0,
  "duplicate_keys": 0,
  "output_file": "metafield_def_audit_<date>.csv"
}

Output Format

CSV file metafield_def_audit_.csv with columns:

definition_id, owner_type, namespace, key, name, type, description_present, metafields_count, pinned, is_unused, is_undocumented, is_duplicate_key, flags

Error Handling

Error	Cause	Recovery
`THROTTLED`	API rate limit exceeded	Wait 2 seconds, retry up to 3 times
`ACCESS_DENIED` for an owner type	Caller lacks the read scope for that resource	Skip that owner type with a warning row in the CSV
`metafieldsCount` returns null	Owner type does not expose count, or count is still computing	Treat as `unknown`; do not flag as `unused`
Owner type not supported in API version	Newer owner type not yet available	Skip with warning; re-run after API version upgrade

Best Practices

Run quarterly and after any app install/uninstall — apps frequently leave behind their definitions when removed.

Do NOT bulk-delete unused definitions without first searching the storefront theme for references to that namespace.key. Theme liquid may read a definition that has zero saved values yet (e.g., a newly added field that has not been populated).

Pin the most-used definitions (pinnedPosition set) to surface them in the merchant Admin UI; un-pinned but heavily used definitions are a UX smell.

Duplicate keys across owner types are not always wrong (e.g., custom.notes on both ORDER and CUSTOMER may be intentional) but they almost always indicate copy-paste creation — review for consistency in type and validations.

Pair with a metafield-value sampling skill (per owner type) before any cleanup to confirm true zero usage; counts can lag in fresh stores.

Keep the CSV in version control alongside theme/app schema docs — the diff over time is the cleanest record of catalog-data evolution.