Skip to main content
Agents in Databricks and Snowflake can use Atlan’s tools too. Add MCP Gateway to the platform as an external MCP server. Its agents can then search your Registry, read artifacts, and call enabled servers their identity can access. Calls are permission-checked and recorded in MCP session activity when the audit write succeeds.
Databricks (AI Playground and agents) and Snowflake (Cortex Agents and Cortex Code) add Atlan MCP Gateway as an MCP server. The gateway provides OAuth sign-in, a permission check, and session audit, and serves Atlan Registry tools plus your registered servers' tools.Databricks (AI Playground and agents) and Snowflake (Cortex Agents and Cortex Code) add Atlan MCP Gateway as an MCP server. The gateway provides OAuth sign-in, a permission check, and session audit, and serves Atlan Registry tools plus your registered servers' tools.
You need permission to register an external MCP service in the platform and workspace access in Atlan. Platform availability and creation privileges depend on your account configuration.
Both platforms use the same Atlan values:

​
Choose a platform and identity

A shared agent connection uses the registered agent’s permissions and audit identity. A connection for each person uses that person’s Atlan permissions. Choose the identity before configuring the platform’s authentication method. These connections let the platform’s agents call Atlan. To let Atlan call a Snowflake or Databricks MCP server instead, use the corresponding template in the MCP integration catalog. Verify. Ask the platform’s agent “Find the Atlan skill that summarizes support tickets.” Confirm it calls registry__search_artifacts, then find the call in Atlan under Tool calls.