pranavlal/JarbasAI — reverse-engineered prompt
Reverse engineered prompt
I want you to get this Jarbas project working as a local, open source voice assistant on Debian, basically a Mycroft style assistant but with the extra experimental AI features this fork talks about. It should run without any cloud pairing or web account setup, use local config files, and support voice input, speech output, hotwords, skills, and the built in client or server style messaging if that part is still usable.
Please keep the rough spirit of the repo, including the bundled skills and the more interesting features like chatbot style responses, image or face related skills, browser automation, music search, and desktop helper behavior, but be realistic because the README says this branch is unstable and deprecated. If something is broken, patch it enough to get a usable demo instead of trying to make every old feature perfect.
Set it up the easiest practical way, preferably in its own virtual environment, with clear run steps and a basic config that works offline where possible. Look up current docs or replacements online if you need to.
Want more depth? Deep Reverse