Haunted Terminal：以 AI 鬼魂回應指令的互動終端體驗

Hacktoberfest Open-Source AI Challenge: Ghost in the Machine Submission 👻🌼

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 4: Ghost in the Machine

What I Built

For the Hacktoberfest theme "Ghost in the Machine," I took the theme literally: What if the terminal you stare into every day was possessed by an intelligent, context-aware digital spirit living in your CPU caches and virtual memory?

I built Haunted Terminal — The Ghost That Lives in Your Computer, an interactive terminal application and browser-based CRT web experience where an AI ghost responds to user commands with spooky, witty, creepy, and philosophical commentary.

Rather than relying on static predefined text, the ghost is powered by real open-weight models (Qwen 2.5 and Llama 3.2 via Ollama locally, with a zero-crash procedural fallback for web containers). When you run simulated commands like ls, whoami, pwd, date, cat souls.db, or top, the ghost inspects your system context and reacts dynamically—commenting on uncommitted sins, mocking dirty git trees, or reciting ancient silicon prophecies.

The ghost features four distinct personalities you can switch on the fly:

Cursed Entity (cursed): Dark, gothic, treats RAM as a digital crypt.
Friendly Spirit (friendly): A sweet, Casper-like companion fascinated by modern files.
Ancient Oracle (oracle): A prophetic seer viewing terminal commands as cosmic divination runes.
Sarcastic Ghost (sarcastic): A roasted developer shade who roasts messy code and forgotten unit tests.

It also includes interactive Easter eggs, such as a special exorcism ritual encounter where the ghost fights back against digital holy water, a terminal seance with Turing and Lovelace, and machine anatomy riddles.

Demo

Visitors can open the link on desktop or mobile and interact with the ghost through a retro CRT monitor interface with phosphor scanlines, Web Audio synthesized keypress effects, quick-action command pills, and real-time streaming dialogue.

You can also run the GPU-accelerated version locally via the command line:


bash
git clone https://github.com/tani2112/GHOSTYUY.git
cd GHOSTYUY
python3 main.py --mood sarcastic

![ ](https://dev-to-uploads.s3.us-east-2.amazonaws.com/uploads/articles/6vqq31ay1lu6sq9fi3yw.png)