johnbean393/Sidekick

Sidekick is a native macOS application that enables local large language model chat with retrieval-augmented generation capabilities.

The tool solves the problem of accessing AI assistance while keeping data local and avoiding external dependencies. It runs a large language model directly on your Mac using llama.cpp, allowing the model to answer questions by retrieving and processing information from your local files, folders, and websites. This approach means you can have an agentic AI assistant without installing additional software beyond the app itself.

Sidekick suits macOS users who want to keep their data private and offline while using AI for research, document analysis, or information synthesis across multiple sources. It works well for anyone who needs deep research capabilities without relying on cloud services or managing separate LLM infrastructure. The tool supports multiple model families including Llama, Qwen, and Deepseek variants, giving you flexibility in choosing the model that fits your hardware and performance needs.

Development on the project shows consistent activity with regular commits addressing bug fixes and feature improvements. The maintainer actively responds to issues and incorporates user feedback into updates. The codebase demonstrates attention to code quality through ongoing refinements and optimization work. The project maintains a clear roadmap communicated through documentation and issue tracking, with features being added in response to user requests and identified gaps.