llm-workflow-engine

Why we included this project

For anyone who does real work with OpenAI models from a terminal, LWE replaces the tab-and-refresh loop of a chat web page with an actual command-line tool. It wraps ChatGPT and GPT-4 into an interactive shell where you can pick up old conversations, adjust model parameters, and then step out into a Python library when you want the same behavior inside your own scripts. The workflow side is the part worth looking at: prompts become reusable templates that chain into multi-step pipelines, and provider plugins extend it to Cohere, Hugging Face, and others. Teams that want one consistent way to prompt and orchestrate LLM calls get a solid base without hand-rolling every API request. Tool use, a plugin architecture, and an experimental Docker image round it out for people who plan to extend it.