Tools

Tool index

Twenty-eight v1 tools: 22 reads, 5 draft-side writes, one publish. Names are versioned (_v1) and stable. Each row links to the detail page that documents the tool's arguments and behavior.

Vocabulary: master, projection, channel

Tool names use Peak PIM's internal words.

A master is the canonical item every store inherits (a product, collection, variant, metafield definition, or media file). A projection is one store's version of that item — the values that store actually publishes. A channel is a connected Shopify store, so channel_id means store id.

All 28 tools

Tool	Scope	What it does
`list_products_v1`	read	Lean product rows with paging/search/sort
`list_collections_v1`	read	Lean collection rows
`list_variants_v1`	read	Lean variant rows
`list_masters_v1`	read	Power-read: full-detail items of any type
`get_master_v1`	read	One item by id
`get_master_with_projections_v1`	read	One item plus its per-store versions
`get_master_related_v1`	read	One item plus related entities
`list_projections_v1`	read	Per-store versions, filterable
`get_projection_v1`	read	One per-store version by id
`update_master_attributes_v1`	write	Patch an item's draft fields, with optional optimistic concurrency
`update_projection_attributes_v1`	write	Patch a store version's draft fields
`publish_master_v1`	publish	Publish an item live to Shopify
`list_markets_v1`	read	Markets with per-store records and catalogs
`get_market_v1`	read	One market by id
`list_metafield_definitions_v1`	read	Custom-field definitions and how each is typed
`list_metaobject_definitions_v1`	read	Structured-content types with field structure and entry count
`list_metaobject_entries_v1`	read	Records under one metaobject definition
`get_master_translations_v1`	read	Every locale a product or collection is translated into, per store
`list_store_sales_channels_v1`	read	A store's Shopify sales channels (live Shopify read)
`get_master_sales_channels_v1`	read	The channels one item is published to on one store (live)
`list_drops_v1`	read	The account's Drops — scheduled change windows
`get_drop_v1`	read	One Drop with every change and its apply/revert status
`list_master_drops_v1`	read	The Drops that will change one item, and when
`get_data_health_v1`	read	Cached catalog-quality state per dimension
`list_data_health_issues_v1`	read	The individual items behind one section's numbers
`run_data_health_scan_v1`	write	Recompute one dimension (non-destructive)
`fill_completeness_gaps_v1`	write	Batch-fill fields that are still empty (drafts)
`reconcile_value_divergence_v1`	write	Batch-harmonize a field across stores (drafts)

Behaviors shared by every tool

Tenancy from the token. No account argument anywhere; cross-account ids answer "not found". See Security & limits.
Data, not instructions. Results are inert entity data; assistants are told not to act on text embedded in them.
Bounded output. page_size is clamped at 100 with a visible [note]; text beyond 20,000 characters is truncated with a marker.
Bounded time. 20 s per call — 120 s for publish and the slow data-health operations.
Drafts first. Every write edits Peak PIM drafts; only publish_master_v1 reaches Shopify. Writes are blocked with 423 during import/refresh reconciliation.
Honest errors. Malformed arguments come back as a tool error with the reason and retry_available: true; conflicts are 409; missing scopes are 403 with re-authorization guidance.