MCP Tools Reference
Once connected, Claude has access to the following tools for working with your Shopify data.
Report Tools
| Tool | Description | Type |
|---|---|---|
report_list | List all available reports. Optionally filter by group name. | Read |
report_get | Get the full details of a specific report by its ID. Returns complete report configuration. | Read |
report_run | Run a report and return the query results as JSON. | Read |
report_create | Create a new report with specified configuration. | Write |
report_update | Update an existing report configuration. | Write |
report_delete | Permanently delete a report. This action cannot be undone. | Delete |
report_list
List all available reports with optional filtering.
Parameters:
group(optional): Filter reports by group name
Returns: List of reports with ID, title, group, and description.
report_get
Get complete report configuration.
Parameters:
report_id(required): The UUID of the report
Returns: Full report configuration including query, visualization settings, and metadata as JSON.
report_run
Execute a report and get results.
Parameters:
report_id(required): The UUID of the report to run
Returns: Query results as JSON with columns and rows.
report_create
Create a new report.
Parameters:
report_json(required): Report configuration as JSON string (TsReport structure with title, query, explore, and fields)
Returns: Created report ID and title.
report_update
Update an existing report.
Parameters:
report_json(required): Report configuration as JSON string including theidfield
Returns: Updated report ID and title.
report_delete
Permanently delete a report.
Parameters:
report_id(required): The UUID of the report to delete
Returns: Confirmation message.
Dashboard Tools
| Tool | Description | Type |
|---|---|---|
dashboard_list | List all available dashboards with ID, title, description, and block count. | Read |
dashboard_get | Get the full details of a specific dashboard including blocks, filters, and metadata. | Read |
dashboard_create | Create a new dashboard with specified blocks and configuration. | Write |
dashboard_update | Update an existing dashboard configuration. | Write |
dashboard_delete | Permanently delete a dashboard. This action cannot be undone. | Delete |
dashboard_list
List all dashboards.
Parameters: None
Returns: List of dashboards with ID, title, description, and block count.
dashboard_get
Get complete dashboard configuration.
Parameters:
dashboard_id(required): The UUID of the dashboard
Returns: Full dashboard configuration including blocks, filters, and metadata as JSON.
dashboard_create
Create a new dashboard.
Parameters:
dashboard_json(required): Dashboard configuration as JSON string (TsDashboard structure with title, blocks, and filters)
Returns: Created dashboard ID and title.
dashboard_update
Update an existing dashboard.
Parameters:
dashboard_json(required): Dashboard configuration as JSON string including theidfield
Returns: Updated dashboard ID and title.
dashboard_delete
Permanently delete a dashboard.
Parameters:
dashboard_id(required): The UUID of the dashboard to delete
Returns: Confirmation message.
Data Model Tools
| Tool | Description | Type |
|---|---|---|
datamodel_summary | Get a compact summary of the schema with all table and column names. | Read |
datamodel_get | Get the full datamodel schema or a specific entity by path. | Read |
datamodel_patch | Apply batch changes to the datamodel schema using JSON Patch operations. | Write |
datamodel_summary
Get a compact overview of available tables and columns.
Parameters: None
Returns: Object with:
explores: List of explore namestables: Map of table names to their labels and column names
Use this to discover available fields for building queries.
datamodel_get
Get full schema or specific entity.
Parameters:
path(optional): Name-based path to get specific entity
Path Examples:
/tables[name=orders]— get single table/tables[name=orders]/columns[name=total]— get single column/tables[name=orders]/datasource— get datasource config/explores[name=sales]— get single explore/explores[name=sales]/relationships[to=customers.id]— get single relationship/crossRelationships[to=refunds.order_id]— get cross relationship
Returns: Full schema (no path) or specific entity as JSON.
datamodel_patch
Apply batch changes to the schema.
Parameters:
operations(required): Array of patch operations
Operations:
add— add new entityreplace— update existing valueremove— delete entity
Path Syntax:
/tables[name=orders]/columns[name=total]/label— target a column field/tables[name=orders]/columns/-— append new column (- means append)/explores[name=sales]/relationships[to=customers]— target a relationship
Examples:
Add a column:
{
"op": "add",
"path": "/tables[name=orders]/columns/-",
"value": {
"name": "discount",
"kind": "measure",
"dataType": "float"
}
}
Update a label:
{
"op": "replace",
"path": "/tables[name=orders]/columns[name=total]/label",
"value": "Order Total"
}
Delete a column:
{
"op": "remove",
"path": "/tables[name=orders]/columns[name=old_field]"
}
ACE Expression Tools
| Tool | Description | Type |
|---|---|---|
ace_validate | Validate an ACE expression. Returns whether the expression is valid and any error message. | Read |
ace_run | Run/execute an ACE expression and return the result. | Read |
ace_validate
Check if an ACE expression is valid.
Parameters:
expression(required): The ACE expression to validate
Returns:
valid: booleanerror: error message (if invalid)
ace_run
Execute an ACE expression.
Parameters:
expression(required): The ACE expression to execute
Returns: Query results as JSON.
Tool Categories Summary
Read-Only Tools (Safe)
These tools only read data and don't modify anything:
report_list,report_get,report_rundashboard_list,dashboard_getdatamodel_summary,datamodel_getace_validate,ace_run
Write Tools (Require Caution)
These tools create or modify data:
report_create,report_updatedashboard_create,dashboard_updatedatamodel_patch
Delete Tools (Destructive)
These tools permanently delete data:
report_deletedashboard_delete
In Claude Web, set read-only tools to Auto-approve and write/delete tools to Needs approval for a safe yet productive workflow.