Claude CLI

A command-line interface for interacting with Claude, Anthropic's AI assistant. Ask questions, process files, pipe command outputs directly to Claude, and execute natural language commands from your terminal.

Prerequisites

• Python 3.7 or higher
• An Anthropic API key (get one here)

Installation

1. Clone or download this repository:

git clone <repository-url>
cd claude-cli

2. Run the installation script:

chmod +x install.sh
./install.sh

3. Source your shell configuration:

source ~/.bashrc  # or source ~/.zshrc for Zsh

4. Set up your API key:

export ANTHROPIC_API_KEY='your-key-here'

Add this to your .bashrc or .zshrc to make it permanent.

The installation script will:

• Create a virtual environment
• Install all dependencies
• Set up the CLI command in your PATH
• Configure your shell environment

Usage

Direct Questions

Ask Claude questions directly:

ai "what is the capital of France?"
ai "how do I list all Python files in a directory?"

Natural Language Command Execution

Execute terminal commands using natural language:

ai -c "list all files recursively in this folder"
ai -c "show disk usage for all drives"
ai -c "find all python files modified in the last week"
ai -c "create a new directory called projects"

The -c or --command flag tells Claude to:

1. Convert your natural language request into a shell command
2. Show you the command it's about to execute
3. Run the command and display its output

Process Files

Process file contents:

cat document.txt | ai summarize
cat code.py | ai "explain this code"

Command Output Processing

Process the output of other commands:

ls -la | ai "explain these files"
ps aux | ai "which processes are using the most CPU?"

Examples

# Get coding help
ai "write a python function to calculate fibonacci numbers"

# Execute system commands in natural language
ai -c "compress all jpg files in current directory"
ai -c "show memory usage"
ai -c "find large files over 1GB"

# Analyze logs
tail -n 50 /var/log/system.log | ai "find any error patterns"

# Explain commands
man grep | ai "explain grep's most useful options"

# Summarize documents
cat meeting_notes.txt | ai "create a bullet-point summary"

Command Execution Safety

When using the -c/--command flag:

• The tool will always show you the command it plans to execute before running it
• Review the command to ensure it matches your intentions
• Be cautious with destructive operations (delete, remove, etc.)
• The tool executes commands with your current user permissions

Environment Variables

• ANTHROPIC_API_KEY: Your Anthropic API key (required)

Technical Details

• Uses Python virtual environment for isolated dependencies
• Installed in ~/bin with automatic PATH configuration
• Dependencies managed through requirements.txt
• Maximum response length: 1000 tokens
• Uses Claude 3.5 Sonnet model
• Natural language command conversion using Claude
• Safe command execution with subprocess module

Error Handling

The script will:

• Check for the presence of the API key
• Provide clear error messages if the API call fails
• Show command execution errors if they occur
• Exit with appropriate status codes

Limitations

• Maximum response length is set to 1000 tokens
• Requires active internet connection
• API usage is subject to Anthropic's rate limits and pricing
• Command execution is limited to your user permissions
• Complex multi-step commands may need to be broken down

Project Structure

claude-cli/
├── README.md
├── requirements.txt
├── install.sh
├── main.py
└── .venv/          # Created during installation

Troubleshooting

If you encounter any issues:

1. Ensure Python 3.7+ is installed: python3 --version
2. Verify the API key is set: echo $ANTHROPIC_API_KEY
3. Check that ~/bin is in your PATH: echo $PATH
4. Try reinstalling: ./install.sh
5. For command execution issues, try running the command directly in terminal

Contributing

Feel free to open issues or submit pull requests for improvements.

License

MIT License - See LICENSE file for details.