llmwizard
Scans your hardware, recommends a model honestly, installs it over SSH with every step shown first, then works as your private chat client.
The story
- Problem
- Running your own AI model means a pile of Linux commands, guesswork about hardware, and security holes for most people.
- Solution
- The app scans the hardware, recommends a model, and shows each install step before running it, with an undo. It's private by default, with opt-in LAN sharing and a firewall rule.
- Result
- Has 140+ automated tests, including screenshot tests, validated against LXD containers and Hyper-V VMs.
Features
- Hardware scan with estimated vs. measured model speed
- Recipe engine: every step previewed, run over SSH, and undoable
- Long installs survive the phone sleeping
- Built-in chat and model management for Ollama
Built with
Support
Questions, bugs or feature requests? Email info@dreamz-code.com and put “llmwizard” in the subject.
Need something like llmwizard?
I build custom apps with the same care. Tell me about your project.