Matrix

A cli-based prompt manager for LLMs. Supports OpenAI (GPT-3, ChatGPT, GPT-4 etc), Anyscale Endpoints, vLLM and any other LLMs that support the OpenAI API.

Features

Provides a simple and easy to use interface for interacting with LLMs
Manages and stores API tokens for multiple providers
Supports multiple LLMs
Manages and stores reusable prompts

Installation

Use pipx to install matrix.

pipx install git+https://github.com/recrsn/matrix.git

To install a specific version, use the @ syntax.

This will install the matrix command line tool in an isolated virtual environment.

Quick Start

matrix providers register openai
Name (leave empty to use alias): OpenAI
URL: https://api.openai.com/v1
Token (leave empty if not required): <your-openai-api-token>
Registered provider OpenAI with alias openai

matrix prompts run --provider openai --model gpt-4-turbo-preview generic --stream
You: Who are you?
🤖 Assistant: I am an AI developed by OpenAI, designed to assist you with a wide range of queries and tasks. Whether you have questions, need information, or require help with something specific, I'm here to provide assistance. How can I help you today?
You: ^D

Concepts

Provider: A provider is a service that provides access to an LLM. For example, OpenAI, Anyscale,etc.
Prompt: A system prompt that is used to start a conversation with an LLM.
Assistant: A binding of a provider and an LLM to a prompt for a specific use case.

Commands

matrix providers: Manage providers
matrix assistants: Manage assistants
matrix prompts: Manage prompts

See matrix --help for more details.

Configuration

Matrix stores its configuration as INI files in - ~/Applicaion Support/matrix/ on MacOS - ~/.config/matrix/ on Linux - %APPDATA%\matrix\ on Windows

License

BSD-2-Clause License

See LICENSE for more details.