Cursor builds the Consulting Chaos minigames

Alvin Leung built Consulting Chaos during a learning day, using ChatGPT to refine the concept and Cursor Agent mode to implement it. He reports a playable Python game after a couple of hours, with feedback from three colleagues.

The concept translated consulting frustrations into four minigames: typing jargon, quick mathematics, fitting meetings onto a grid and escaping a maze. ChatGPT helped shape the initial specification. Cursor then created a simple scene-based architecture so each game could be implemented separately and connected to a common loop.

Leung supplied requests for individual mechanics, including a typing game with penalties. The agent generated logic, state management, interface rendering, input handling and scoring. Follow-up requests adjusted difficulty and penalty behavior. The owner retained the creative decisions and tuned details such as enemy counts and puzzle blocks.

The finished prototype used Python 3.13, tkinter and JSON persistence for high scores. The reported result is a specific playable artifact, not an independently benchmarked development-speed claim. The account provides a demonstration and code link, but the small group’s favorable feedback does not establish broad usability or cross-platform reliability. Human design and gameplay refinement remained part of the process.

Prompt
Original prompt

Source: medium.com · Alvin Leung