ChainFury is a production-grade chaining engine that powers TuneChat and Tune Studio. It provides a Python library (chainfury) and server (chainfury_server) for composing LLM-based workflows, including built-in support for Retrieval Augmented Generation (RAG), multi-model backends, and prompt engineering. The project enables developers to build and self-host conversational AI applications by chaining together LLM calls with retrieval and other tooling.