I built a haunted museum where dead tech (and my old repos) talk back 👻
This is a submission for the Hacktoberfest Open-Source AI Challenge Week 4: Ghost in the Machine What I Built The Haunted Museum of Dead Tech 👻 is a museum where the technologies we buried come back as ghosts, and they still have a lot to say. You walk through the halls, pick an exhibit and talk to its ghost. Each one tells you its story, defends itself from its critics and tries to convince you…
This submission for the Hacktoberfest Open-Source AI Challenge demonstrates the Haunted Museum of Dead Tech, an interactive exhibition where historically obsolete technologies "return as ghosts." Eight ghosts - the Floppy Disk, VHS, Walkman, Game Boy, Encarta, Internet Explorer, Clippy, and MSN Messenger - are present, each shaped like its object and dressed in pastel colors.
Visitors can engage in conversation with the ghosts, who narrate their stories, defend themselves against critics and plead their case to return from the dead.
Various elements are included in the museum such as an altar for offerings, a guestbook for visitors to leave messages, a debate hall for two ghosts to argue live, and a personal graveyard where visitors can commemorate their own GitHub repositories that they have stopped contributing to.
Each ghost is powered by a Gemma 3 (4B) model running locally using Ollama, ensuring no API keys or cloud dependencies. The museum's architecture comprises Fastify (Node) for the museum server, Astro with plain CSS for the interface, and Markdown files for each ghost's history. New ghosts can be added by simply creating a new Markdown file, and the model is instructed to remember its history vaguely rather than inventing specific dates.
Visitors can converse with the ghosts in their preferred language, as the system adapts accordingly. The Haunted Museum of Dead Tech operates offline, preserving personal memories and experiences and ensuring no server or internet connection is required for the experience.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
