A minimal bash utility that converts natural language queries into shell commands using a local Ollama instance.

• Ollama installed and running on your machine
  • Install from: https://ollama.ai
  • Pull a model: ollama pull devstral-2-small (or any other model you prefer)

1. Clone this repository:

git clone <repository-url>
cd qq

2. Run the installation script:

./install.sh

3. Reload your shell:

source ~/.bashrc  # or ~/.zshrc for zsh users

Simply type qq followed by your natural language request:

qq for each file in this directory, print the first line to stdout

This will automatically generate and populate your shell with:

find . -maxdepth 1 -type f -exec head -n 1 {} \;

1. Shell Detection: qq automatically detects your shell (bash, zsh, or fish)
2. Query Processing: Sends your natural language request to Ollama with shell-specific context
3. Command Generation: Ollama generates the appropriate shell command
4. Auto-population:
   • zsh: Command appears directly in your prompt buffer (ready to execute)
   • bash: Command is added to history (press Up arrow to access)
   • fish: Command appears in your command line buffer
   • Other shells: Command is printed to stdout

By default, qq uses the devstral-2-small model. To use a different model, set the QQ_MODEL environment variable:

export QQ_MODEL=codellama  # If you want to use the `codellama` model, add this to your ~/.bashrc or ~/.zshrc

For best results with shell commands:

• devstral-2-small (default, fast and efficient)
• codellama (optimized for code)
• llama3.2 (fast and accurate)
• llama3.1 (larger, more capable)

Pull a model with:

ollama pull devstral-2-small

# File operations
qq list all python files modified in the last 7 days

# Text processing
qq count the number of lines in all txt files

# System information
qq show disk usage sorted by size

# Git operations
qq show all commits from the last week

# Process management
qq find and kill all node processes

• Be specific in your requests for better results
• The generated command appears instantly (1-2 seconds)
• Review generated commands before executing them
• For complex operations, you can refine the natural language query

# Check if Ollama is installed
which ollama

# Install Ollama if needed
# Visit https://ollama.ai

# Use a smaller, faster model
export QQ_MODEL=devstral-2-small

# Ensure Ollama is running
ollama list

• Make sure you've reloaded your shell after installation
• For bash, use the Up arrow to access the generated command from history
• Check that the installation added the source line to your RC file

Remove the source line from your shell RC file:

# Edit ~/.bashrc or ~/.zshrc and remove:
source "/path/to/qq/qq.sh"

MIT

Contributions welcome! Please open an issue or pull request.