Gemini CLI Integration Manual (Model Context Protocol)

This section guides you through connecting the Gemini CLI to the platform using the Model Context Protocol (MCP). Connecting your client allows a terminal agent to leverage a large context window over your entire project toolchain to analyze requirements, track tasks, and audit code repositories.

Gemini CLI Click to enlarge

Prerequisites & Access Token Generation

Before configuring the CLI, you must generate a secure access token. This token serves as your REST API key and authenticates your MCP session.

Navigate to the MCP — Gemini CLI dashboard.
Click the blue Create token button located in the upper-right corner of the instruction panel.
Copy the generated token immediately.

Important Security Note

The token configuration snippet fills in automatically upon creation and is only displayed once. Do not commit your active token to a shared repository. If you are using a per-project configuration file within a shared repository, prefer using a global file in your home directory instead.

Configuration Setup#

Gemini CLI reads its server profiles from a JSON configuration file. You can define this configuration globally or on a per-project basis.

Configuration File Paths

Global Configuration: ~/.gemini/settings.json
Per-Project Configuration: ./.gemini/settings.json

JSON Configuration Block#

Add the following block to your settings.json file, replacing tf_YOUR_TOKEN with the access token you generated:

Gemini CLI configures MCP servers in ~/.gemini/settings.json  (global) or .gemini/settings.json  (per project) and supports remote HTTP servers via httpUrl.

Step 1#

Add this to the settings file:

json

{
  "mcpServers": {
    "Think4Ever": {
      "httpUrl": "https://cell-ssm-use1-0005.us.portal.think4ever.com/mcp",
      "headers": { "Authorization": "Bearer tf_YOUR_TOKEN" }
    }
  }
}

Step 2#

Restart Gemini CLI and run /mcp to confirm the Think4Ever server is connected and its tools are listed.

Step 3#

Use /mcp desc to see tool descriptions, and mention Think4Ever in your prompt so the model reaches for the right tools.