MindQuest

Screenshot of the first Quest of 'The Broken Crown'

Inspiration

We have a passion for computer games that leave room for everyone's own imagination to comprehend the stories in their own way. That is why we opted for a text-based adventure game. We didn't want to do 'yet another text-based game'. The underlying focus on mental health is important. This game allows people from all cultures and backgrounds struggling with mental health issues to get invested in a story where they can relate to the main character, as the character struggles with similar mental health issues. Currently, there is only 1 story implemented, focusing on self-worth. However, the idea is to have multiple stories, each focusing on a different mental health issue, where the user can select the story they want to experience.

What it does

Allows the player to interact with an environment, items, and characters in a form of text inputs. The text inputs are not hard-coded. The player can write whatever they like. The input is analysed by a large language model (LLM), which maps the player input to any of the available actions and allows the story to progress further without needing to specify an exact command. Also, some of the responses are generated by an LLM too.

How we built it

Python backend interacting with Google's Gemini LLM model, running a Flask server for HTTP request commucation between backend and frontend. Frontend is built with JavaScript + React. Figma was used to help sketch out and structure the story.

Challenges we ran into

MongoDB Atlas was planned to be used for game state, but we had issues establishing the connection. Instead, SQLite and hard-coded paths were used to achieve a working state in the shortest possible time.

Accomplishments that we're proud of

Achieved to implement various APIs, set up a server so the backend and frontend can communicate with each other even though it's built using different languages. Managed to create a decent story, considering none of us have any experience with storytelling.

What we learned

inter-language (programming) communication, interacting with LLMs through an API, using LLM programatically to control the flow of execution, building dynamically loaded webpages using React, to stay awake for more than 24 hours