ai-agents-pixels
A browser-based pixel-art simulation where six AI agents walk around a virtual trading room, run trades, and update a shared portfolio, purely for fun, no real money involved.
PixelTrade is a browser-based simulation of a trading room drawn in pixel-art style. Six AI agents walk around a virtual office, visiting workstations, running trades, and updating a shared portfolio, all rendered in retro pixel graphics. It is a visual simulation rather than a real trading tool; no actual money or markets are involved.
The dashboard shows the portfolio balance, profit and loss, and an equity chart that updates as the simulation runs. You can click on any of the eleven stations in the room to send the nearest available agent there. The simulation can run at normal speed or at two or four times normal speed. A settings panel lets you toggle autopilot on and off, adjust animation, and set an aggression level for the agents, which controls how often they trade versus how often they idle.
The README is written in Thai, so readers who do not read Thai will need a translation for some details. The core behavior is described in the structure: each agent cycles through idle, walking, and working states. When working, the agent produces an outcome (a trade, an analysis note, etc.) that affects the portfolio. Trades have roughly a 66% win rate with randomized profit and loss amounts.
The project uses React 18 loaded directly from a CDN and Babel running in the browser to handle JSX, so there is no build step and no package manager setup required. Opening the index.html file in a browser is enough to run it. Everything runs client-side with no backend or server needed.
The code is organized into a handful of JSX files handling the room rendering, simulation logic, sidebar panel, trade history, and analysis view. It is licensed under MIT.
Where it fits
- Open index.html in a browser to watch six pixel-art AI agents simulate a trading room with a live portfolio and equity chart.
- Modify agent aggression settings to explore how trading frequency affects simulated portfolio performance.
- Use as a visual template for building your own multi-agent browser simulation with pixel art graphics.