Ask CLI

Ask CLI is a Linux shell LLM assistant designed to assist system administrators with their daily tasks. It connects to local or internet-accessible Large Language Models (LLMs) using APIs compatible with Openai api standards, allowing users to send queries and receive concise, actionable responses directly in the terminal.

It works seamlessly with tools like llama.cpp and Ollama, which enable running all the most popular open LLMs, such as LLaMA 3, Gemma, and Mistral, providing flexibility and enhanced privacy.

Example Usage

You can ask for specific command-line tasks:

ask How can I compress a directory into a tar.gz file?

Output:

tar -czvf archive.tar.gz /path/to/directory

Features

• Communicates with local LLM servers (e.g., Ollama or Llama.cpp).
• Supports public LLM APIs (e.g., OpenAI's GPT models).
• Customizable configuration via YAML.
• Streamed responses for real-time interaction.
• Lightweight and easy to use.

Installation

Option 1: Using the Installation Script (Recommended)

To install Ask CLI with the provided script, run the following command:

curl -sfL https://github.com/kmkamyk/ask-cli/raw/main/install.sh | sh -

This will automatically download and install ask.py to /usr/bin/ask and config.yml to /etc/ask/config.yml.

To uninstall, run:

curl -sfL https://github.com/kmkamyk/ask-cli/raw/main/install.sh | sh -s uninstall

Option 2: Manual Installation (Using Git)

1. Clone the repository:

   git clone https://github.com/kmkamyk/ask-cli.git
   cd ask-cli

2. Install requirements for Python.

   python3 -m pip install -r requirements.txt

3. Copy the ask.py script to /usr/bin/ and make it executable:

   sudo cp ask.py /usr/bin/ask
   sudo chmod +x /usr/bin/ask

4. Place the config.yml configuration file in /etc/ask/:

   sudo mkdir -p /etc/ask
   sudo cp config.yml /etc/ask/config.yml

5. Place the configuration file in /etc/ask/config.yml.

Example use.

To use Ask CLI, follow these steps:

1. Install a Local LLM Server
   Install a compatible LLM server such as Ollama. Ollama provides an easy way to run large language models locally.

2. Download a Model
   After installing Ollama, download a model that you want to use. For example, to download and run the llama3.1 model, use the following command:

   ollama run llama3.1

3. Configure Ask CLI
   Edit the /etc/ask/config.yml configuration file to point to the LLM server. If Ask CLI and the LLM server (e.g., Ollama) are running on the same machine, use localhost as the IP address. Otherwise, specify the IP address of the machine hosting the LLM server.

   Example configuration:

   api:
     base_url: "http://localhost:11434/v1"  # Replace localhost with the server's IP if needed
     api_key: "sk-no-key-required"
   model:
     name: "llama3.1"  # The name of the model to use
     system_prompt: "You are a helpful CLI assistant. Only output the command and nothing else."
     temperature: 0.0

4. Start Asking
   Once the configuration is set up, you can use the ask command to send queries to the LLM. For example:

   ask How to find the oldest files in the current directory?

Configuration

The tool reads its configuration from /etc/ask/config.yml. Example:

api:
  base_url: "http://localhost:8080/v1"
  api_key: "sk-no-key-required"

model:
  name: "Meta-Llama-3.1"
  system_prompt: "You are a helpful CLI assistant in BASH. Display only the command on the screen and nothing else."
  temperature: 0.7