johnbean393/Sidekick

A native macOS app that allows users to chat with a local LLM, responding with information from files, folders, and websites on your Mac, ensuring privacy and offline functionality.

Core Features

Local-first LLM chat with Retrieval Augmented Generation (RAG) over personal data (files, folders, websites).

Advanced AI capabilities including web search, function calling, and a dedicated Deep Research agent.

Privacy-focused design: all conversations happen offline, and your data remains secure on your device.

Built-in llama.cpp backend for GGUF models, with optional compatibility for OpenAI compatible APIs.

GitHub Repo Documentation

Detailed Introduction

Sidekick is a native macOS application designed for private, local-first AI interactions. It leverages llama.cpp to run large language models directly on your Mac, allowing users to query their personal files, folders, and web content without data leaving their device. With RAG, web search, function calling, and a dedicated Deep Research agent, Sidekick transforms your Mac into a powerful, secure, and context-aware AI assistant for research, productivity, and complex task execution.