Open EPM ← Back to openepm.dev

MCP Overview and Guidelines

An MCP client connects to Open EPM so it can discover and call permitted operations. Start with a functionality module below to learn its workflow, the access it requires, and what counts as a completed result.

From a user request to an Open EPM operation

  1. Sign in through the client and check the verified session identity. Use the application router to check enrollment and discover the workspaces and applications visible to your account.
  2. Choose the application using its stable workspace and application IDs. Discover the endpoint's callable tools and the function catalog for that scope.
  3. Translate the intended action into the discovered contract. A tool such as read_v2_program takes a typed program. Its function, such as facts.cube.scan, selects the application operation.
  4. For a write, use the preview and submission path required by that function. Keep the exact typed request, any required key, and the durable result identity for recovery.

Current Open EPM servers also offer the versioned Multidimensional Cube Harness as MCP resources. You can review it, retrieve it for a client-side installation you authorize, or continue with the tools already available to you. Published skill reference

Use the prompt library for reusable user instructions, the tool reference for callable contracts, and the client walkthrough for a first bounded read and reviewed write.

In this section

Discussion

Discussing MCP Overview and Guidelines

The discussion service requires JavaScript and a connection to Firebase.