Run a private AI assistant on your own PC that can open files and apps without sending data to the cloud.
Automate repetitive desktop tasks like cleaning up a downloads folder using a shareable skill.
Let an AI browse the web and click through pages for you while you watch it work.
Build and share custom skills that teach the assistant new workflows.
| 0pen-sourcer/hearth | 13127905/deep-learning-based-air-gesture-text-recognition- | 42kyynfqjv-dot/deepseekradio | |
|---|---|---|---|
| Stars | 15 | 15 | 15 |
| Language | Python | Python | Python |
| Setup difficulty | moderate | moderate | hard |
| Complexity | 3/5 | 3/5 | 4/5 |
| Audience | vibe coder | developer | developer |
Figures from each repo's GitHub metadata at analysis time.
Easiest on Windows via installer, source install on macOS or Linux needs Python 3.11 or newer.
Hearth is a local AI assistant that runs on your own computer instead of in the cloud. It talks, listens, opens your apps, reads and writes your files, controls a real web browser you can watch, and remembers things about you across sessions. Nothing you do gets sent anywhere else, except a web search, and only when you ask for one. There is no account to create and no telemetry collecting your data. The project describes itself as different from three common kinds of AI tools. Chat interfaces like LibreChat or Open WebUI just let you talk to a model, they cannot touch your files or take actions. Coding agents like Aider or Cline are built for writing code in a project folder, not for running your whole computer. Cloud assistants like ChatGPT or Claude are powerful but depend on a company's servers, rules, and pricing. Hearth aims to be a fourth option, an assistant that lives on your machine and can actually operate it, including your files, shell, apps, browser, and desktop clicks and typing. The framework itself is called Hearth, and it ships with a default assistant named JARVIS, which you can rename to anything. You can extend what it can do with skills, shareable folders that teach it a new workflow, such as cleaning up a downloads folder or turning a batch of photos into a contact sheet. You can reach Hearth through a terminal command line tool, a desktop and web app, a headless bridge, or an MCP server connection. Installation is easiest on Windows through a downloadable installer, offered in a Full version that bundles its own model server for GPU users, or a Lite version for people who already run a local model server like LM Studio or Ollama. It can also be installed from source on Windows, macOS, or Linux using Python 3.11 or newer. Hearth can connect to a locally running model server automatically, or you can optionally point it at a cloud provider such as Gemini, OpenAI, Grok, or OpenRouter while keeping your files, voice, and memory local. The project notes it is at an early preview stage, with the command line tool and desktop app being the most reliable parts right now, while voice features are still being polished.
A local AI assistant that runs on your own PC, no cloud needed, that can open apps, edit files, browse the web, and remember you.
Mainly Python. The stack also includes Python, llama.cpp, PowerShell.
Use freely for any purpose, including commercial use, as long as you keep the copyright notice.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly vibe coder.
This repo across BitVibe Labs
double-check against the repo, no cap.