A multi-API provider support version based on gemini-cli, maintaining full compatibility with gemini-cli. Users can seamlessly switch between different large language model services. I named it gemini-cli-nexus.

nexus means "connection point" or "core", this name emphasizes the tool's role as a central hub for interacting with various AI large models.

This name was recommended by Gemini 😄

Supported models include:

• Google Gemini (default)
• OpenAI GPT series
• Anthropic Claude series
• DeepSeek
• Kimi2
• Other models compatible with the above APIs

• Automatically detects available API keys from environment variables
• Intelligently selects the optimal provider
• Supports provider fallback mechanism

• All providers use the same calling interface
• No need to modify existing code when switching providers
• Maintains backward compatibility

Please ensure you have Node.js 20 or higher installed before installation.

The installation method is consistent with gemini-cli. It's recommended to use npx in the project directory.

$ npx https://github.com/ai-embedded/gemini-cli-nexus

Note: If using npx, please set environment variables before running gemini-cli-nexus. If environment variables are not set, gemini-cli-nexus will use the default gemini model.

For global installation, first ensure gemini-cli is not installed. If gemini-cli is already installed, please uninstall it first:

$ sudo npm uninstall -g gemini-cli
$ sudo npm install -g https://github.com/ai-embedded/gemini-cli-nexus

When starting gemini-cli-nexus, it will automatically detect API keys in environment variables and .env files, and automatically select providers based on available keys. After startup, you'll have 2 more login options compared to gemini-cli. Select different providers to use different API services.

How would you like to authenticate for this project?

● 1. Login with Google
  2. Use Gemini API Key
  3. Vertex AI
  4. Use OpenAI API Key
  5. Use Anthropic API Key

export OPENAI_API_KEY="sk-your-openai-key-here"
export OPENAI_BASE_URL=https://api.openai.com/v1 
# or
export ANTHROPIC_API_KEY="sk-ant-your-anthropic-key-here"
export ANTHROPIC_BASE_URL=https://api.anthropic.com

# Set model (unified configuration method)
export MODEL="gpt-4o-mini"

Create a .env file in the project root directory:

# Option 1: OpenAI
OPENAI_API_KEY=sk-your-openai-key-here
OPENAI_BASE_URL=https://api.openai.com/v1  # Optional, defaults to OpenAI

# Option 2: Anthropic
ANTHROPIC_API_KEY=sk-ant-your-anthropic-key-here
ANTHROPIC_BASE_URL=https://api.anthropic.com  # Optional, defaults to Anthropic

# Option 3: Gemini (original/default)
# GEMINI_API_KEY=AIza-your-gemini-key-here

# Option 4: Vertex AI (Google Cloud)
# GOOGLE_CLOUD_PROJECT=your-project-id
# GOOGLE_CLOUD_LOCATION=us-central1
# GOOGLE_API_KEY=your-google-api-key  # Optional for express mode

# Model selection (optional)
# MODEL=gpt-4o-mini
# MODEL=claude-3-5-sonnet-20241022
# MODEL=gemini-2.5-pro
MODEL=claude-sonnect-4-20250514

You can refer to the .env.example file in the project directory for configuration. Please be careful not to commit .env to avoid key leakage.

If you need to re-select authentication method:

# Method 1: Delete authentication configuration
rm ~/.gemini/settings.json

# Method 2: Edit settings file, delete selectedAuthType field
# Edit ~/.gemini/settings.json

# Method 3: Complete reset
rm -rf ~/.gemini/

Or use the /logout command after logging into the command line to log out and log in again.

For detailed usage instructions of gemini-cli, please refer to gemini-cli

• qwen-code

This project is developed based on gemini-cli. Thanks to gemini-cli.

Developed based on the original gemini-cli, following the same license.