Side Quest: A Local AI Game Master That Sends You Outside
This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass What I Built Side Quest is a tiny game where an AI running on your own laptop gives you one small real-world mission, and the only way to win is to go outside. Open it on your phone. Choose how long you have (10, 20, 45 or 90 minutes) and your mood (curious, calm, with kids, need to clear my head…). A local…
Side Quest is a compact game that utilizes an AI running on your personal computer to generate modest, real-world tasks. The only way to complete the game is by venturing outdoors. After launching the game on your device, you select the duration of the task (10, 20, 45 or 90 minutes) and your current mood (curious, calm, with children, or needing a mental break).
A local Gemma model, acting as the game master, creates a unique challenge tailored to your current time, season, and weather conditions. For instance, on a scorching October morning in Arizona, the game master might prompt you to locate the most vibrantly colored autumn leaf and trace its veins with your finger. Upon completion, you snap a photo as evidence, which is then reviewed by a vision model.
This model checks if the photo meets the specified criteria and records it in your field journal, which keeps track of quests completed and time spent outdoors. The entire process typically takes around 30 seconds. Side Quest is designed for individuals who want to step outside but struggle to initiate the process. It targets those who need a mission, those taking a break to clear their heads, or friends playing a lighthearted challenge together.
The quests focus on observation tasks such as finding a heart-shaped leaf, capturing a reflection in a puddle, or identifying three types of bark. They do not involve physical exercise or require photographing strangers. The game's codebase is relatively small, consisting of approximately 180 lines of Python and one HTML file. It relies solely on standard libraries, with no additional installations required.
The game employs open-weight models from Ollama for both the game master and the referee. The game master generates whimsical quests, while the referee validates the submitted photos against predefined success criteria. The game adapts to the user's location, season, and live weather data, all sourced from Open-Meteo. The only limitation is that the complete end-to-end deployment requires the application to be hosted online for mobile devices.
However, the current version functions well on desktop computers via a web-powered application. Data privacy is a priority, as your photos and location remain stored on your device, not transmitted to any external servers.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
