The Room That Forgot You: A Museum Mystery with an Archivist That Runs in Your Browser
This is a submission for the Hacktoberfest Open-Source AI Challenge Week 4: Ghost in the Machine . What I Built The museum knows every donor's name. It cannot tell you who washed the cups. The Room That Forgot You is a first-person browser mystery set in a railway museum after closing. Its archivist remembers important visitors and celebrated events. Someone else survives only in the margins: an…
The Room That Forgot You is a first-person browser mystery set in a railway museum after closing hours. An archivist in the game remembers key visitors and celebrated events, while other details are left in the margins. Players enter the museum using a flashlight and a handheld receiver, with the goal of giving that overlooked person a place in the record.
The main mechanic involves changing the room by connecting evidence. Finding a document is just the beginning, as players must then link it to another record, establish what happened, and confirm the connection. As the player advances, distorted spaces regain their sounds, shape, and meaning.
The game consists of three galleries, eighteen records, and nine connections. In the "Departures" gallery, players see that someone abandoned the station, returned time and time again, and the station sheltered people during a flood. Notable townspeople carried out important work, keeping the doors open. As the player discovers additional records and makes connections, the room transforms, revealing hidden truths.
The game's architecture relies on React 19, React Three Fiber, Three.js, TypeScript, and Vite to render a procedural museum. Two open-weight models, MiniLM and Qwen2.5-1.5B-Instruct, work together to process questions, match evidence to conclusions, and validate answers. These models follow the MIT license, and the game's own code and narrative are released under the same license.
Players can access the game at any time, as no account is required. It uses a keyboard and mouse for navigation, with WASD/arrow keys for movement and Shift for rushing. The mouse is used for looking around, flashing, and inspecting objects. The game saves progress within the browser, and the models can be downloaded for testing.
The game snapshot, developed in commit 64ad8d6, has passed 21 unit tests and a production build. Development evaluations included 29/30 checks, accounting for evidence validation, save recovery, controls, and encounter logic. The mixed pipeline response time averaged around 12 milliseconds, primarily due to neural retrieval.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
