This is a submission for the Hacktoberfest Open-Source AI Challenge Week 4: Ghost in the Machine. Haunted Terminal AI — The Forgotten Signal is an open-source, AI-powered, text-based mystery adventure built with Python and Google FLAN-T5 Small.
What if an abandoned computer suddenly started talking to you?
The year is 2026. A forgotten computer wakes up with a mysterious message:
Find my memories before the signal fades.
Players enter an abandoned building, explore five interconnected locations, discover hidden clues, and communicate with a mysterious AI called ECHO.
The challenge theme, Ghost in the Machine, inspired me to create a story about technology, forgotten memories, and an AI whose past has become a mystery.
You can play the game directly in Google Colab:
To get started:
help to see the available commands.
No local installation is required. An internet connection is needed to load the model.
The complete project is publicly available on GitHub:
The repository includes the Python game in a Jupyter notebook and a README with gameplay instructions.
The project is open for exploration, feedback, and improvements.
I used the following technologies:
The game combines traditional Python programming with AI-generated dialogue.
Python handles the game structure, including locations, hidden clues, player commands, and endings. The AI model is used to generate responses when the player asks ECHO questions.
This approach allowed me to experiment with AI storytelling while keeping the main gameplay logic predictable.
Players can use commands such as look, explore, go ROOM, ask QUESTION, clues, map, status, and unlock ECHO.
I tested the game in Google Colab and successfully reached:
GOOD ENDING — THE AWAKENING
My test results were:
The game handled room navigation, clue discovery, and the final terminal interaction.
One limitation I discovered was that the small language model sometimes generated responses that did not fit the mystery's storyline.
This was an important lesson: integrating an AI model is only the beginning. Creating consistent and meaningful AI dialogue requires careful prompting, testing, and improvements.
Using an openly available AI model allowed me to experiment with AI-powered storytelling without depending on a paid proprietary API. I could inspect the code, modify the prompts, test the model, and understand how the different parts worked together.
Open-source tools also make learning more accessible. Other developers can study the project, experiment with the game, and suggest improvements.
For me, open innovation is about learning through experimentation and sharing what we create. I would like to improve Haunted Terminal AI by:
Building this project helped me learn more about Python, open-weight language models, debugging, and interactive storytelling.
Sometimes, the most interesting stories begin with a forgotten signal. 👻
Thanks for checking out my project! Feedback and suggestions are welcome.