Powered by Zoomin Software. For more details please contactZoomin

Semaphore 5.13 MCP Server Guide

Prerequisites and Server Configuration

  • Last Updated: November 20, 2026
  • 3 minute read
    • Semaphore
    • Version 5.13
    • Documentation

Before connecting an MCP client or AI agent to Semaphore, make sure your environment meets the requirements below.

The Semaphore MCP Server is available starting with Semaphore 5.13.0. It provides access to supported Knowledge Model Management (KMM) operations through MCP-compatible clients.

Before you begin

Make sure you have:

  • Access to Semaphore 5.13.0 or later with the MCP Server available.
  • The MCP Server URL for your deployment environment.
  • Network connectivity between your MCP client and the MCP Server.
  • A user account with permissions for the KMM operations you intend to perform.
  • A valid authentication token:
    • A Semaphore access token for on-premises deployments.
    • A JWT token for Progress Data Cloud deployments.
  • An MCP client that supports HTTP transport, custom HTTP headers, and MCP tool discovery.

For instructions on obtaining a token, see Authentication and Permissions.

Find your MCP Server endpoint

For on-premises deployments, the MCP Server endpoint follows this format:

http://<semaphore_studio_url>/ai/mcp

Replace <semaphore_studio_url> with the hostname and port of your Semaphore Studio instance.

For example, if Semaphore Studio is available at:

http://semaphore.example.com:5080

the MCP Server endpoint is:

http://semaphore.example.com:5080/ai/mcp

For Progress Data Cloud deployments, use the MCP Server endpoint configured for your cloud environment.

Use the protocol, hostname, and port configured for your environment. Use HTTPS where supported by your deployment to protect credentials and data in transit.

Check your client requirements

Your MCP client must be able to:

  • Connect to an MCP server over HTTP.
  • Send custom HTTP authentication headers.
  • Discover and invoke the tools exposed by the server.

The authentication header depends on the deployment environment:

  • On-premises: X-Api-Key with a Semaphore access token.
  • Progress Data Cloud: Authorization with a Bearer JWT token.

The Semaphore MCP Server uses token-based authentication rather than OAuth.

For instructions on generating tokens and configuring authentication, see Authentication and Permissions.

Choose an MCP client

You can connect using an MCP-compatible client that supports the requirements above.

Visual Studio Code

VS Code supports HTTP MCP servers with custom headers.

A connection to an on-premises Semaphore MCP Server was verified using VS Code with GitHub Copilot Chat in Agent mode. After configuring the X-Api-Key header, the server connected successfully and exposed its tools for discovery and use.

For an example of the VS Code configuration, see Authentication and Permissions.

GitHub Copilot CLI

GitHub Copilot CLI can be configured to connect to the Semaphore MCP Server using an MCP configuration file:

~/.copilot/mcp-config.json

For configuration instructions, see Connecting GitHub Copilot CLI.

Note: VS Code with GitHub Copilot Chat and GitHub Copilot CLI are separate MCP clients. Their configuration procedures differ.

Verify your connection

After configuring your MCP client:

  1. Confirm that the MCP Server endpoint is accessible from the machine running your MCP client.
  2. Check that the MCP Server URL ends with /ai/mcp.
  3. Confirm that your client sends a valid authentication token in the HTTP header required for your deployment environment.
  4. Start the MCP server connection in your client.
  5. Verify that the connection is active and the server's tools are available.

In VS Code, check that the MCP server status is Running and that the client has discovered the available tools.

In GitHub Copilot CLI, use /mcp to inspect the configured server connection.

To confirm that the connection can retrieve Semaphore data, ask your MCP client to list the available knowledge models.

A successful response returns model identifiers and names accessible to the authenticated user. The results depend on the models and permissions in your environment.

If the connection fails, see Limitations and Troubleshooting.

Alert