An anime waifu on your Linux desktop, powered by local LLMs.
Maidserver is a simple launcher for running a playful local AI companion with llama.cpp.
It does not require an API key. It does not require a cloud account. It does not ship a giant model inside the repo.
Clone it, install the tools, run the script, and talk to your local desktop companion.
Maidserver gives you a ready-to-run local character assistant using:
llama.cpp- GGUF models from Hugging Face
- a tuned system prompt
- simple repeat-control settings
- a clean Linux-first launch script
The goal is simple:
A local AI companion should be easy to start, fun to use, and light enough for ordinary Linux users to try.
- Linux
gitcmake- C/C++ build tools
- internet connection for first model download
On Debian/Ubuntu:
sudo apt update
sudo apt install -y git cmake build-essentialcd ~
git clone https://github.com/ggml-org/llama.cpp
cd llama.cpp
cmake -B build
cmake --build build --config Release -jAfter building, verify:
~/llama.cpp/build/bin/llama-cli --helpgit clone https://github.com/thanks-cohn/Maidserver.git
cd Maidserver
chmod +x maidserver.sh./maidserver.shThe first run may download the model through llama.cpp.
Maidserver currently uses:
mradermacher/Impish_LLAMA_4B-GGUF
Impish_LLAMA_4B.Q8_0.gguf
The model is not included in this repository.
Maidserver is not trying to reinvent local AI.
It is trying to make local AI feel approachable.
Linux already has powerful tools. Maidserver adds a friendly face to that power: a small, personal, character-driven assistant that runs locally and can grow into desktop integration, memory, voice, scripting, and automation.
The future of personal AI should not belong only to cloud dashboards.
It should also live on the user’s own machine.
- One-command installer
- Local model cache
- Desktop launcher
- Optional GUI
- Optional voice mode
- Persistent character memory
- Multiple character presets
MIT