Let your AI tools act on SELECT, not just ask it questions

SELECT’s MCP server now connects to the API, so Claude Code, Cursor, Codex and other MCP clients can update insights, log actions, build dashboards and manage monitors for you.

When we launched SELECT Copilot over MCP (Model Context Protocol), your AI client could tell you which warehouse to resize or which query to rewrite but it couldn't do anything about it. You needed to come back to SELECT to update the insight and log the action by hand.

SELECT's MCP server now connects to the SELECT API, the same API the SELECT app runs on. Your AI client can read and change what's in SELECT: insights, actions, comments, dashboards, monitors, budgets, usage groups, Jira ticket links, automated savings and more. For the full list of what's exposed, see the API docs.

Examples of what you can ask

• "Update the insight and log an action for the Snowflake_WAREHOUSE resizing change."
• "Add a block to our Home dashboard showing Databricks spend by workspace for the last 90 days."
• "Leave a comment on this budget explaining last week's overrun."

It acts as you

Every call runs with your own SELECT sign-in and permissions. Your agent can't see or change anything you couldn't in the SELECT app, and there's no API key to create or store. AI clients that ask before making changes will still ask you first.

Get started

If you're already connected, there's nothing to set up. The new capabilities show up the next time your client connects to SELECT. Otherwise, the MCP Server docs take a couple of minutes.

Please try out these new capabilities and share your feedback with us.