MCP Tool Permissions

Which permissions each MCP tool requires. Use this reference when scoping a service account’s group for MCP use.

Each MCP tool requires one or more permissions. Service accounts inherit their group’s permissions, so to use a tool the group must include all of its required permissions. Tools that require permissions the service account lacks are automatically hidden from the MCP client.

See Custom groups for how to set up a group, and Service accounts for how to issue an API key.

This page lists the MCP tools available in the current Datafold app (us.datafold.com, eub.datafold.com, and dedicated or customer-hosted deployments). For Datafold Classic, see the Classic MCP tool permissions.

​

Minimum permissions to enable every MCP tool

To give a service account access to every MCP tool except those that require Admin, create a custom group with the permissions below and assign your service account to it.

• Cancel diffs (cancel_datadiff)
• Create diffs (create_datadiff)
• Edit monitors (edit_monitor)
• List data sources (list_data_sources)
• List users (list_users)
• Run monitors (run_monitor)
• View diffs (view_datadiff)
• View monitors (view_monitor)

Custom groups cannot hold the Admin permission. Tools that require Admin are available only to accounts in the built-in admin group. Add the service account to admin only if it needs those tools.

​

Tools by category

​

Organization

Tool	What it does	Required permissions
list_org_members	Retrieves all active members of the authenticated user’s organization.	List users (list_users)

​

Data Sources

Tool	What it does	Required permissions
get_dataset_schema	Get the column schema for a table on a data source.	List data sources (list_data_sources)
list_data_sources	Retrieves all data sources accessible to the authenticated user.	List data sources (list_data_sources)
run_query	Executes a SQL query against the specified data source and returns the results.	List data sources (list_data_sources), Create diffs (create_datadiff)

​

Source Repos

Tool	What it does	Required permissions
list_source_repos	List git source repositories configured on the org.	Admin (admin)

​

Data Diffs

Tool	What it does	Required permissions
cancel_datadiff	Cancel a running or queued data diff.	Cancel diffs (cancel_datadiff), View diffs (view_datadiff)
create_datadiff	Launches a new data diff to compare two datasets (tables or queries).	Create diffs (create_datadiff), View diffs (view_datadiff)
get_datadiff_overview	Retrieves a structured overview of a data diff, mirroring the UI’s Overview tab.	View diffs (view_datadiff)
get_datadiff_result_section	Retrieves detailed results for a specific section of a data diff, corresponding to UI tabs.	View diffs (view_datadiff)
list_datadiffs	Lists existing data diffs for the organization, ordered by creation date (newest first).	View diffs (view_datadiff)

​

Monitors

Monitors in the current Datafold app are data diff monitors.

Tool	What it does	Required permissions
list_monitors	Lists the diff monitors you can see, newest first.	View monitors (view_monitor)
get_monitor	Get one diff monitor, including its schedule and data diff definition.	View monitors (view_monitor)
get_monitor_as_code	Export one monitor as a monitors-as-code YAML fragment.	View monitors (view_monitor)
get_monitor_run_results	Get a monitor’s runs, newest first, with each run’s state (ok, alert, error, or cancelled).	View monitors (view_monitor)
get_monitors_schema	Returns the JSON Schema of the monitors-as-code YAML format (.datafold/monitors.yml).	View monitors (view_monitor)
provision_monitors	Create, update, or adopt diff monitors from a monitors-as-code YAML config.	Edit monitors (edit_monitor)
trigger_monitor_run	Start a check of one monitor now, outside its schedule.	Run monitors (run_monitor)
trigger_monitors_run	Start checks for many monitors at once, selected by monitor IDs, tags, or both.	Run monitors (run_monitor)

​

Help and feedback

Tool	What it does	Required permissions
get_datafold_guide	How-to guides for the tools on this server: workflows, parameters, limits, and results.	—
submit_feedback	Submit feedback to the Datafold team — report bugs, request features, or share general thoughts.	—