Sidekick
A native macOS app for chatting with a local large language model that can answer using files, folders and websites on the Mac, entirely offline, with an optional bring-your-own-key mode.
Sidekick is a native macOS app that lets a user chat with a local large language model that can respond using information from files, folders and websites on the Mac, without installing any other software. It runs entirely offline by default, using a built-in llama.cpp inference engine, and supports modern GGUF models such as Qwen3.5 out of the box.
Retrieval-augmented generation lets a user configure collections of files, folders and websites, called experts, around a specific area of interest; activating an expert lets Sidekick fetch and cite relevant material, including direct quotes with page numbers that open the source in a viewer, and files can also be dragged directly into the input field. A web search option lets it use current information, and users can bring their own API key for OpenAI-compatible remote providers, including OpenAI, Anthropic, Google AI Studio, DeepSeek, Groq, MiniMax, Mistral and xAI, while keeping a primarily local-first workflow.
It is written in Swift and released under the MIT license.