Before installing CCS, make sure you're logged into Claude CLI with your subscription account:

claude /login

macOS / Linux / Windows

npm install -g @kai/ccs

Compatible with npm, yarn, pnpm, and bun package managers.

macOS / Linux

curl -fsSL ccs.kaitran.ca/install | bash

Windows PowerShell

irm ccs.kaitran.ca/install | iex

# Use Claude subscription (default)
ccs "Review this architecture design"

# Switch to GLM for cost-optimized tasks
ccs glm "Create a simple REST API"

# Use GLM for all subsequent commands until switched back
ccs glm
ccs "Debug this issue"
ccs "Write unit tests"

All major package managers are supported:

# npm (default)
npm install -g @kai/ccs

# yarn
yarn global add @kai/ccs

# pnpm (70% less disk space)
pnpm add -g @kai/ccs

# bun (30x faster)
bun add -g @kai/ccs

~/.ccs/config.json:

{
  "profiles": {
    "glm": "~/.ccs/glm.settings.json",
    "default": "~/.claude/settings.json"
  }
}

If Claude CLI is installed in a non-standard location (D drive, custom directory), set CCS_CLAUDE_PATH:

export CCS_CLAUDE_PATH="/path/to/claude"              # Unix
$env:CCS_CLAUDE_PATH = "D:\Tools\Claude\claude.exe"   # Windows

See Troubleshooting Guide for detailed setup instructions.

You have both Claude subscription and GLM Coding Plan. Two scenarios happen every day:

1. Rate Limits Hit: Claude stops mid-project → you manually edit ~/.claude/settings.json
2. Cost Waste: Simple tasks use expensive Claude → GLM would work fine

Manual switching breaks your flow. CCS fixes it instantly.

The Solution:

ccs           # Use Claude subscription (default)
ccs glm       # Switch to GLM fallback
# Hit rate limit? Switch instantly:
ccs glm       # Continue working with GLM

One command. Zero downtime. No file editing. Right model, right task.

graph LR
    subgraph "User Command"
        CMD[ccs glm]
    end

    subgraph "CCS Processing"
        CONFIG[Read ~/.ccs/config.json]
        LOOKUP[Lookup profile → settings file]
        VALIDATE[Validate file exists]
    end

    subgraph "Claude CLI"
        EXEC[claude --settings file_path]
    end

    subgraph "API Response"
        API[Claude Sub or GLM API]
    end

    CMD --> CONFIG
    CONFIG --> LOOKUP
    LOOKUP --> VALIDATE
    VALIDATE --> EXEC
    EXEC --> API

• One Command: ccs glm to switch to GLM, ccs to use Claude subscription - no config file editing
• Smart Detection: Automatically uses right model for each task
• Persistent: Switch stays active until changed again

• No Downtime: Switching happens instantly between commands
• Context Preservation: Your workflow remains uninterrupted
• Seamless Integration: Works exactly like native Claude CLI

🚧 Work in Progress: This feature is experimental and not fully tested. Use with caution.

CCS includes intelligent task delegation via the /ccs meta-command:

Install CCS commands:

ccs --install    # Install /ccs command to Claude CLI

Use task delegation:

# After running ccs --install, you can use:
/ccs glm /plan "add user authentication"
/ccs glm /code "implement auth endpoints"
/ccs glm /ask "explain this error"

Remove when not needed:

ccs --uninstall  # Remove /ccs command from Claude CLI

Benefits:

• ✅ Save tokens by delegating simple tasks to cheaper models
• ✅ Use right model for each task automatically
• ✅ Seamless integration with existing workflows
• ✅ Clean installation and removal when needed

ccs              # Use Claude subscription (default)
ccs glm          # Use GLM fallback
ccs --version    # Show CCS version and install location
ccs --install    # Install CCS commands and skills to ~/.claude/
ccs --uninstall  # Remove CCS commands and skills from ~/.claude/

macOS / Linux:

curl -fsSL ccs.kaitran.ca/uninstall | bash

Windows PowerShell:

irm ccs.kaitran.ca/uninstall | iex

• YAGNI: No features "just in case"
• KISS: Simple bash, no complexity
• DRY: One source of truth (config)

Complete documentation in docs/:

• Installation Guide
• Configuration
• Usage Examples
• Troubleshooting
• Contributing

We welcome contributions! Please see our Contributing Guide for details.

CCS is licensed under the MIT License.