Before you start
- Gemini CLI installed (run
npm install -g @google/gemini-cliif not already). - A Brivvy account on a plan that includes MCP access.
- A text editor for modifying
settings.json.
Steps
-
Open a terminal and navigate to the Gemini CLI config directory. The path depends on your platform:
Create the file if it does not exist.
Platform Path macOS / Linux ~/.gemini/settings.jsonWindows %APPDATA%\gemini\settings.json -
Open
settings.jsonin a text editor. The file uses standard JSON format. If the file is empty, start with an opening and closing curly brace{}as the base structure. -
Paste the following configuration inside the top-level object. If a
mcpServersblock already exists, add"brivvy"as an additional entry rather than creating a second block. - Save the file. Make sure the JSON is valid before closing the editor. A trailing comma after the last entry or mismatched brackets are common issues that will prevent Gemini CLI from starting.
-
Run
geminiin the terminal. The CLI readssettings.jsonon startup and attempts to connect to each configured MCP server. -
Inside the Gemini CLI session, run
/mcp auth brivvyto start the OAuth flow. Your browser will open an authorization page where you can sign in to Brivvy and grant access. -
After authentication completes, type
/mcpand press Enter. The output lists all connected servers and their available tools.brivvyshould appear with tools such asget_voiceandlist_templates. - Send a prompt that requires brand voice context, for example: “Using Brivvy, get the brand voice rules and write a two-sentence product update.” Gemini CLI will call the appropriate Brivvy tool and incorporate the returned guidelines into the response.
How to know it worked
The/mcp command shows brivvy with a connected status and a list of tools. A test prompt that references Brivvy returns a response shaped by the brand voice rules rather than a generic output.