Repolode
Submit
johnbean393

johnbean393/Sidekick

A native macOS app for chatting with a local LLM that answers using your files, folders and websites, with no extra software needed.

Open-source alternative to:ChatGPTClaude

View on GitHubWebsite
Stars
3,323
Forks
146
Open issues
35
Last commit
last month

What is it?

Sidekick is a native macOS chat app that runs a local large language model through a built-in llama.cpp backend, so conversations can stay on your Mac. You organize files, folders and websites into experts, and when an expert is active, Sidekick retrieves relevant passages with RAG (retrieval-augmented generation) to answer your questions and cite its sources.

Beyond local models, Sidekick can connect to OpenAI-compatible APIs using your own key, with presets for providers such as OpenAI, Anthropic, Google AI Studio and DeepSeek. It also includes function calling for actions like drafting emails in your default mail client, web search, a Deep Research agent that reads many webpages and writes a report, and a memory feature that keeps preferences across conversations.

The app adds a Canvas for editing text and code, LaTeX and chart rendering for responses, and image generation that requires macOS 15.2 or later with Apple Intelligence. It requires a Mac with Apple Silicon and at least 8 GB of RAM.

Key features

  • 01Chat with local GGUF models through a built-in llama.cpp backend
  • 02Organize files, folders and websites into experts for RAG-based answers with sources
  • 03Connect OpenAI-compatible APIs with your own key, including presets for major providers
  • 04Function calling for actions such as drafting emails and running calculations
  • 05Deep Research agent for multi-step web research that produces a report
  • 06Canvas for creating and selectively editing text and code

Use cases

  • Searching and quoting from your own research papers or documents offline
  • Asking questions about a project folder without uploading files to the cloud
  • Drafting emails or summarizing web research on a Mac with Apple Silicon
  • Comparing local models with remote APIs in one chat interface

Quick start

  1. 1

    Install via Homebrew

    brew install --cask arcadi4/tap/sidekick-ai
  2. 2

    Follow the getting started guide to set up the app

Things to know

  • Requires a Mac with Apple Silicon and at least 8 GB RAM.
  • Image generation needs macOS 15.2 or later and Apple Intelligence.
  • Remote models require your own API key from a provider.

In collections

Private AI knowledge bases and RAGOpen-source AI agent frameworks & appsRun LLMs locallyOpen-source Mac appsAI chatbots & chat clients

Topics

#macos#local-llm#rag#llama-cpp#swift#chatbot#function-calling#deep-research#agentic-ai#agents

FAQ

What is Sidekick?
A native macOS app for chatting with a local LLM that answers using your files, folders and websites, with no extra software needed. Sidekick is a native macOS chat app that runs a local large language model through a built-in llama.cpp backend, so conversations can stay on your Mac. You organize files, folders and websites into experts, and when an expert is active, Sidekick retrieves relevant passages with RAG (retrieval-augmented generation) to answer your questions and cite its sources.
Is Sidekick free to use? What license does it have?
Sidekick is open source under the MIT license. Check the license terms before using it in commercial products.
Is Sidekick still maintained?
Updated within 6 months. The last commit to Sidekick was on Sep 30, 2026.
What can I use Sidekick for?
Searching and quoting from your own research papers or documents offline Asking questions about a project folder without uploading files to the cloud Drafting emails or summarizing web research on a Mac with Apple Silicon Comparing local models with remote APIs in one chat interface
How do I install Sidekick?
Install via Homebrew: brew install --cask arcadi4/tap/sidekick-ai Follow the getting started guide to set up the app
Where can I get Sidekick?
The source code and installation instructions are on GitHub: https://github.com/johnbean393/Sidekick

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects