What I Built
What if the terminal you use every day had a ghost living inside it?
For Hacktoberfest Week 4, I built SéanceOS — The Ghost in the Machine, a spooky developer experience powered by Gemma 3 4B, an open-weight AI model.
I interpreted the theme as a meeting between the past and present of computing. The people and ideas that shaped technology may be gone, but their influence is still alive in the machines we use today.
SéanceOS turns that idea into an interactive experience with three main parts:
- Possessed Terminal: A terminal-inspired interface where commands receive eerie AI commentary.
- Séance Chamber: A conversation space where users can interact with historical computing pioneers such as Ada Lovelace.
- Día de los Muertos Ofrenda: A digital altar where users can create tributes and generate Calaverita-style poems.
The interface is inspired by old CRT monitors, green phosphor displays, and the mysterious atmosphere of Halloween.
Demo
SéanceOS is designed to run locally.
GitHub repository:
https://github.com/anuuu2507/seance-os
To run the project:
git clone https://github.com/anuuu2507/seance-os.git
cd seance-os
pip install -r requirements.txt
python run_web.py
Make sure Ollama is installed and the Gemma 3 model configured for the project is available locally.
Then open:
http://localhost:8000
Code
The complete source code is available here:
The repository includes the FastAPI backend, frontend, CLI, AI integration, security utilities, and automated tests.
How I Built It
SéanceOS is built around Gemma 3 4B, an open-weight model accessed through Ollama.
The application combines:
- Gemma 3 4B + Ollama for AI generation
- Python + FastAPI for the backend
- HTML, CSS, and JavaScript for the CRT-inspired interface
- Server-Sent Events (SSE) for streaming responses in the Séance Chamber
- Python CLI for an interactive terminal experience
The AI is not just an extra feature added to the interface. It is what gives the terminal its personality and allows the Séance Chamber to respond to the user.
The backend connects the interface to the model, while the frontend presents the experience through the retro CRT design.
The project also includes a security layer designed to block dangerous commands and sanitize sensitive output.
Why Does Open Innovation Matter?
Open innovation was important because I wanted to experiment with an AI experience that feels personal, creative, and connected to the machine it runs on.
Using an open-weight model through Ollama gave me the opportunity to integrate AI inference into a local application instead of depending entirely on a remote, closed API.
That made the project a learning experience as well as a creative one. I could explore how a model is connected to a backend, how prompts shape its responses, and how the same model can be given a completely different personality through the application around it.
Open models also give developers more room to experiment. They make it possible to learn, adapt, and build without making every part of the experience dependent on a hosted service.
For SéanceOS, the open approach supports the idea behind the project:
The ghost lives inside the machine.
The AI experience is built around a model that can run locally, and the project remains open for others to explore, modify, and extend.
My Agent Session
The Spirit of It
The moment that made the idea feel real was asking Ada Lovelace about the soul of computation.
The terminal stopped feeling like just a place to execute commands. It became a place to have a conversation with the history of computing.
That is the feeling I wanted to capture with SéanceOS: technology can feel mysterious, but it is also shaped by human imagination, curiosity, and ideas that continue to live on.
The machines change. The ideas remain.
Top comments (0)