Claude Mode

A Node.js CLI launcher for Claude Code that supports multiple AI providers and models.

Features

• Interactive menu - Guided step-by-step selection of provider, model, and mode
• Quick mode - Command-line arguments for fast launching
• Multiple providers - OpenRouter, Ollama Cloud, Ollama Local, Ollama Custom
• Custom providers - Define your own providers via config file
• Dynamic model discovery - Ollama models discovered automatically via API
• Offline model cache - Falls back to cached models when API is unreachable
• Health checks - Provider availability shown in interactive mode
• Headless mode - Single prompt execution for scripting/automation
• Terminal mode - Interactive Claude Code session
• Shell completions - Tab completion for bash, zsh, and fish

Installation

npm Package (Recommended)

# Install globally
npm install -g @cbuk100011/claude-mode

From Source

# Clone the repository
git clone https://github.com/Cam10001110101/claude-mode-cli.git
cd claude-mode-cli

# Install dependencies
npm install

# Build
npm run build

# Link globally
npm link

Usage

Interactive Menu

Run without arguments to get a guided menu with provider health checks:

claude-mode

This will:

1. Check availability of all providers
2. Prompt you to select a provider
3. Select a model (dynamically discovered for Ollama)
4. Choose terminal or headless mode
5. Enter a prompt (if headless)

Quick Mode

Specify provider, model, and optionally mode/prompt:

# Interactive terminal session
claude-mode openrouter sonnet
claude-mode ollama-local qwen3

# Headless mode with default provider/model
claude-mode -p "list files"

# Headless mode with specified provider/model
claude-mode ollama-local qwen3 h "list all TypeScript files"
claude-mode ollama-cloud glm47 headless "explain this code"

# Using full provider names
claude-mode ollama-custom gpt120 h "run tests"

Commands

claude-mode --help              # Show usage
claude-mode --list              # List all available models
claude-mode list                # Same as above
claude-mode health              # Check provider availability
claude-mode config show         # Show current configuration
claude-mode config init         # Create config file
claude-mode completion bash     # Generate bash completions
claude-mode completion zsh      # Generate zsh completions
claude-mode completion fish     # Generate fish completions

Providers

| Provider | Description | Environment Variable | |----------|-------------|---------------------| | openrouter | OpenRouter API | ANTHROPIC_BASE_URL | | ollama-cloud | Ollama Cloud | OLLAMA_HOST | | ollama-local | Local Ollama | OLLAMA_BASE_URL_LOCAL | | ollama-custom | Custom Ollama server | OLLAMA_BASE_URL_CUSTOM |

Provider Shortcuts

| Shortcut | Provider | |----------|----------| | or, open | openrouter | | oc, cloud | ollama-cloud | | ol, local | ollama-local | | custom, remote | ollama-custom |

Models

OpenRouter (Static)

| Shortcut | Model ID | |----------|----------| | opus | anthropic/claude-opus-4.5 | | sonnet | anthropic/claude-sonnet-4.5 | | haiku | anthropic/claude-haiku-4.5 | | gpt52 | openai/gpt-5.2 | | gpt52-pro | openai/gpt-5.2-pro | | gpt120 | @preset/gpt-oss-120b-cerebras | | glm47 | @preset/cerebras-glm-4-7-cerebras | | deepseek | deepseek/deepseek-v3.2 | | gemini-pro | google/gemini-3-pro-preview | | gemini-flash | google/gemini-3-flash-preview |

Ollama (Dynamic)

Ollama models are discovered automatically via the /v1/models API endpoint. Use claude-mode --list to see currently available models.

Models are cached locally and will work offline if previously discovered.

Configuration

Environment Variables

Create a .env file in your project directory:

# OpenRouter
ANTHROPIC_AUTH_TOKEN=sk-or-v1-your-openrouter-key
# or
OPEN_ROUTER_API_KEY=sk-or-v1-your-openrouter-key

# Ollama Cloud
OLLAMA_HOST=https://ollama.com
OLLAMA_API_KEY=your-ollama-cloud-key

# Ollama Local (default: http://localhost:11434)
OLLAMA_BASE_URL_LOCAL=http://localhost:11434

# Ollama Custom (your own server)
OLLAMA_BASE_URL_CUSTOM=http://192.168.1.100:11434

Config File

Create a config file for persistent settings:

claude-mode config init

This creates ~/.claude-mode/claude-mode.json:

{
  "defaultProvider": "",
  "defaultModel": "",
  "modelDiscoveryTimeout": 5000,
  "healthCheckTimeout": 2000,
  "cacheTTL": 30000,
  "customProviders": [],
  "skipHealthCheck": false,
  "offlineMode": false
}

Custom Providers

Add your own providers in the config file:

{
  "customProviders": [
    {
      "key": "my-server",
      "name": "My Server",
      "baseUrl": "http://my-server:8080",
      "authEnvVar": "MY_SERVER_API_KEY",
      "description": "My custom LLM server"
    }
  ]
}

Mode Options

| Mode | Aliases | Description | |------|---------|-------------| | Terminal | terminal, t, interactive, i | Interactive Claude Code session | | Headless | headless, h, prompt, p | Single prompt execution |

Shell Completions

Add tab completion to your shell:

Bash

# Add to ~/.bashrc
eval "$(claude-mode completion bash)"

Zsh

# Add to ~/.zshrc
eval "$(claude-mode completion zsh)"

Fish

# Save to completions directory
claude-mode completion fish > ~/.config/fish/completions/claude-mode.fish

Examples

# Start interactive menu with health checks
claude-mode

# Quick start with Claude Sonnet (OpenRouter)
claude-mode openrouter sonnet

# Use local Qwen3 for a quick task
claude-mode ollama-local qwen3 h "list files in src/"

# Use Ollama Cloud GLM for code explanation
claude-mode ollama-cloud glm47 h "explain the main function"

# Use remote server with large model
claude-mode ollama-custom gpt120

# Use shortcuts
claude-mode or sonnet h "tell me a joke"
claude-mode ol qwen3

# Check which providers are available
claude-mode health

# List available models
claude-mode --list

# Skip permission prompts (use with caution)
claude-mode --dangerously-skip-permissions

Headless Mode Tools

In headless mode, the following tools are enabled by default:

• Read - Read files
• Edit - Edit files
• Write - Write files
• Bash - Execute bash commands
• Glob - File pattern matching
• Grep - Search file contents

Development

# Install dependencies
npm install

# Development mode (watch)
npm run dev

# Build
npm run build

# Run tests
npm test

# Watch tests
npm run test:watch

# Type check
npm run typecheck

Project Structure

claude-mode-cli/
├── src/
│   ├── index.ts      # Main CLI entry point
│   ├── providers.ts  # Provider and model configurations
│   ├── config.ts     # Configuration management
│   ├── errors.ts     # Error handling utilities
│   ├── *.test.ts     # Test files
├── dist/             # Built output
├── package.json
├── tsconfig.json
├── .env.example
└── README.md

Troubleshooting

Model not found

Ensure the model is available on the selected provider. Use --list to see available models.

Authentication errors

Check that your API keys are correctly set in the .env file or environment variables.

Local Ollama not responding

Ensure Ollama is running: ollama serve

Claude not found

Ensure Claude Code CLI is installed and available in your PATH:

npm install -g @anthropic-ai/claude-code

Provider health check failing

Run claude-mode health to see which providers are available and any error details.

Timeout errors

Increase modelDiscoveryTimeout or healthCheckTimeout in your config file:

claude-mode config init
# Then edit ~/.claude-mode/claude-mode.json

Roadmap

• Secure key storage
• Additional support for providers/models
• Additional support for Codex, Gemini CLI, etc.

License

MIT