Shell-AI: let AI write your shell commands

Shell-AI (shai) is a CLI utility that brings the power of natural language understanding to your command line. Simply input what you want to do in natural language, and shai will suggest single-line commands that achieve your intent. Under the hood, Shell-AI leverages the LangChain for LLM use and builds on the excellent InquirerPy for the interactive CLI.

Installation

You can install Shell-AI directly from PyPI using pip:

pip install shell-ai

Note that on Linux, Python 3.10 or later is required.

After installation, you can invoke the utility using the shai command.

Usage

To use Shell-AI, open your terminal and type:

shai run terraform dry run thingy

Shell-AI will then suggest 3 commands to fulfill your request:

• terraform plan
• terraform plan -input=false
• terraform plan

Features

• Natural Language Input: Describe what you want to do in plain English (or other supported languages).
• Command Suggestions: Get single-line command suggestions that accomplish what you asked for.
• Cross-Platform: Works on Linux, macOS, and Windows.
• Azure Compatibility: Shell-AI now supports Azure OpenAI deployments.

Configuration

Shell-AI can be configured through environment variables or a config file located at ~/.config/shell-ai/config.json (Linux/MacOS) or %APPDATA%\shell-ai\config.json (Windows).

Environment Variables

• OPENAI_API_KEY: (Required) Your OpenAI API key, leave empty if you use ollama
• OPENAI_MODEL: The OpenAI model to use (default: "gpt-3.5-turbo")
• OPENAI_API_BASE: The OpenAI API / OpenAI compatible API endpoint to use (default: None)
• GROQ_API_KEY: (Required if using Groq) Your Groq API key
• SHAI_SUGGESTION_COUNT: Number of suggestions to generate (default: 3)
• SHAI_SKIP_CONFIRM: Skip command confirmation when set to "true"
• SHAI_SKIP_HISTORY: Skip writing to shell history when set to "true"
• SHAI_API_PROVIDER: Choose between "openai", "ollama", "azure", or "groq" (default: "groq")
• SHAI_TEMPERATURE: Controls randomness in the output (default: 0.05). Lower values (e.g., 0.05) make output more focused and deterministic, while higher values (e.g., 0.7) make it more creative and varied.
• CTX: Enable context mode when set to "true" (Note: outputs will be sent to the API)
• OLLAMA_MODEL: The Ollama model to use (default: "phi3.5")
• OLLAMA_API_BASE: The Ollama endpoint to use (default: "http://localhost:11434/v1/")

Config File Example

{
  "OPENAI_API_KEY": "your_openai_api_key_here",
  "OPENAI_MODEL": "gpt-3.5-turbo",
  "SHAI_SUGGESTION_COUNT": "3",
  "CTX": true
}

Config Example for OpenAI compatible

{
   "SHAI_API_PROVIDER": "openai",
   "OPENAI_API_KEY": "deepseek_api_key",
   "OPENAI_API_BASE": "https://api.deepseek.com",
   "OPENAI_MODEL": "deepseek-chat",
   "SHAI_SUGGESTION_COUNT": "3",
   "SHAI_SUGGESTION_COUNT": "3",
   "CTX": true
}

Config example for MistralAI

{
   "SHAI_API_PROVIDER": "mistral",
   "MISTRAL_API_KEY": "mistral_api_key",
   "MISTRAL_API_BASE": "https://api.mistral.ai/v1",
   "MISTRAL_MODEL": "codestral-2508",
   "SHAI_SUGGESTION_COUNT": "3",
   "CTX": true
}

Config Example for Ollama

   {
   "OPENAI_API_KEY":"",
   "SHAI_SUGGESTION_COUNT": "3",
   "SHAI_API_PROVIDER": "ollama",
   "OLLAMA_MODEL": "phi3.5",
   "OLLAMA_API_BASE": "http://localhost:11434/v1/",
   "SHAI_TEMPERATURE": "0.05"
   }

The application will read from this file if it exists, overriding any existing environment variables.

Run the application after setting these configurations.

Using with Groq

To use Shell AI with Groq:

1. Get your API key from Groq
2. Set the following environment variables:

   export SHAI_API_PROVIDER=groq
   export GROQ_API_KEY=your_api_key_here
   export GROQ_MODEL=llama-3.3-70b-versatile
   

Contributing

This implementation can be made much smarter! Contribute your ideas as Pull Requests and make AI Shell better for everyone.

Contributions are welcome! Please read the CONTRIBUTING.md for guidelines.

License

Shell-AI is licensed under the MIT License. See LICENSE for details.