gpt-cli - Enhance command-line interactions with diverse chat LLMs through a simple interface

Introduction to gpt-cli

The gpt-cli is an innovative command-line interface designed to facilitate interaction with large language models (LLMs). This tool offers a seamless way to communicate with advanced AI models directly from your terminal. Ideal for both technical users and those new to command-line environments, gpt-cli emphasizes simplicity and functionality.

Supported Providers

The gpt-cli supports a spectrum of leading AI providers, including:

OpenAI: Access the renowned ChatGPT and other models.
Anthropic: Interact with Claude, another sophisticated AI model.
Google Gemini: Leverage Google's advanced AI capabilities.
Cohere: Utilize Cohere's unique language models.
Other Compatible APIs: Support for APIs compatible with OpenAI, such as Together, OpenRouter, and even local models via LM Studio.

A broad compatibility makes gpt-cli a versatile tool for diverse applications.

Key Features

gpt-cli is equipped with several powerful features:

Command-Line Interface: Engage with AI models like ChatGPT directly within your terminal, providing a straightforward interaction method without needing a graphical interface.
Model Customization: Users have the ability to override default settings such as the model type, temperature, and top_p. This allows for customizing the AI's response behavior according to specific needs.
Usage Tracking: Keep track of your API interactions with comprehensive token counts and pricing details.
Keyboard Shortcuts: Facilitate efficient use with shortcuts like Ctrl-C to clear, Ctrl-D to exit, and Ctrl-R to regenerate responses.
Multi-Line Input: Use multi-line mode for complex queries which require more detailed inputs.
Markdown Support: Control markdown formatting during chat sessions, enhancing the readability of responses according to user preferences.
Predefined Messages: Set predefined messages for custom assistants to set the tone or pre-establish context for conversations.
Multiple Assistants: Swiftly switch back and forth between different assistants, whether they're pre-defined general and developer assistants or your custom configurations.
Flexible Configuration: Tailor your experience via a YAML configuration file that specifies assistant behaviors, model parameters, and API keys.

Installation

Installation is straightforward, especially for Linux/OSX users with Python and pip installed:

pip install gpt-command-line

Alternatively, install the latest version from the source:

pip install git+https://github.com/kharvd/gpt-cli.git

Or by cloning the git repository:

git clone https://github.com/kharvd/gpt-cli.git
cd gpt-cli
pip install .

After installation, add the OpenAI API key to your .bashrc or .zshrc file to activate functionality immediately.

Usage

Starting a conversation with gpt-cli is as simple as typing gpt in your terminal. For example, the command can specify different modes such as general, dev, or bash to tailor responses according to context, like programming or command-line scripting.

A variety of optional arguments enable precise control of the session, allowing the user to:

Disable markdown.
Change models or adjust parameters like temperature.
Log interactions and adjust verbosity.
Input prompts directly or in multi-line form.

Exit the session with :q or Ctrl-D, clear conversations with :c or Ctrl-C, and regenerate last responses with :r or Ctrl-R.

Configuration

The .config/gpt-cli/gpt.yml file is where users can customize their assistants. It allows setting defaults for the entire tool as well as for each assistant, defining models, message roles, and more.

An example configuration to set a default assistant and model might look like:

default_assistant: dev
markdown: True
openai_api_key: <your_open_ai_key>
assistants:
  pirate:
    model: gpt-4
    temperature: 1.0
    messages:
      - {role: system, content: "You are a pirate."}

Additional Chat Bots

gpt-cli also supports interaction with models from other popular AI services through specific setup, including Anthropic's Claude, Google Gemini, and Cohere. Simply add the respective API keys to your environment or configuration file to start using these models.

The gpt-cli stands out as a flexible, powerful tool for interacting with cutting-edge AI models directly from your terminal—a must-have for tech enthusiasts and professionals looking to incorporate AI into their workflows.