Skip to main content

Gemini CLI Integration Guide

9/23/25About 2 min

Gemini CLI Integration Guide

A complete guide to connecting to CCProxy using a custom version of Gemini CLI.

Gemini CLI is a command-line interface for programming agents. The original Gemini CLI can only connect to Google's own Gemini models. We need to use a custom fork, llxprt-code, to enable it to connect to any model proxied by CCProxy.

📦 1. Get the Custom Version

Since the official Gemini CLI repository only supports Google's Gemini models, we need to use a fork that supports connecting to other models. This guide uses llxprt-code as an example, but you can use any other fork you prefer.

🖥️ Install llxprt-code

First, make sure you have Node.js 20 or higher installed on your system.

You can install it via npm:

npm install -g @vybestack/llxprt-code

Or execute it directly via npx:

npx https://github.com/acoliver/llxprt-code

🔌 2. Connect to llxprt

📋 Prerequisites

Before you begin, ensure you have completed the following steps:

  1. Installed Chatspeed
  2. Added the necessary models according to the Quick Start
  3. Completed the CCProxy Basic Configuration
  4. Added the required proxy models
  5. Configured the proxy keys

For llxprt, the model is specified by the user, so configuring a Proxy Group is not a necessary condition.

📝 Configure llxprt

  1. In your command line, switch to your project directory and run llxprt to enter.
  2. After entering llxprt, execute the following configuration commands in order:
    • /provider openai
    • /baseurl http://127.0.0.1:11434/compat_mode/v1
    • /key {ProxyKey}
    • /model {ModelName}

Where:

  • /baseurl http://127.0.0.1:11434/compat_mode/v1: We set this to tool compatibility mode, which can better extend the model's capabilities. If you want to use native tool calling, set the address to http://127.0.0.1:11434/v1.
  • /model {ModelName}: Specifies the model name to use, for example qwen3-coder. You can get the model name from the Chatspeed settings by clicking through the menu: Proxy -> Proxy Services, then find and copy the name of your preferred model.
  • /key {ProxyKey}: You can get the proxy key from the Chatspeed settings by clicking through the menu: Proxy -> Proxy Access Keys. You can create a dedicated key for llxprt or use an existing one.
  1. After the above settings, llxprt can be used normally.
  2. If you want to switch models during use, simply execute /model to select from the list of available models (if there are multiple).

It should look something like this:

╭──────────────────╮
│  > who are you?  │
╰──────────────────╯

✦ I'm an interactive CLI agent specializing in software engineering tasks. I'm designed to help you with coding, debugging, project management, and other
  development-related activities. I can read and write files, execute shell commands, search content, and manage tasks - all within your project environment.

  My primary goal is to assist you safely and efficiently while strictly following the project's conventions and your instructions. I work directly within your
  project directory and can help with tasks like:

   - Understanding and modifying code
   - Running tests and build processes
   - Managing dependencies
   - Searching and refactoring code
   - Creating new files and features

  How can I help you with your current project?