Race to Treasure Island: Building a Web-Native Phaser Game with Free AI
Why the fourth browser game went Phaser instead of Pygbag, how boost/wind/overtaking mechanics evolved, and what the AI model workflow looked like.
Field Notes
Notes from the Pirate Arcade build: tools, decisions, failures, and findings from experimenting with zero-cost AI-assisted development.
Why the fourth browser game went Phaser instead of Pygbag, how boost/wind/overtaking mechanics evolved, and what the AI model workflow looked like.
All three browser-playable games got an art-direction pass: elegant ship paddles, a treasure-barge longboat, fortress-styled bricks, and automated testing to protect them. Here's what was done and what still needs work.
Three arcade games are now browser-playable via Pygbag. Can the same pipeline handle a turn-based board game with AI opponents, save/load, and property trading? Here's the assessment.
Treasure Cove (Breakout) is now playable in the browser. The second port took a fraction of the first — most of the infrastructure was already in place.
Cannonball Clash is now playable in the browser. Here's what it took to get a Python+Pygame game running on the web, what broke, what worked, and what's next.
Evaluating strategies for bringing Pirate Arcade games to the browser: Pygbag/WebAssembly with minimal code changes vs full TypeScript/Canvas rewrite.
How Pirate Arcade v2.0.0 ships to users with zero cloud spend — GitHub free tier for CI and releases, Cloudflare Pages for the site, and a local workstation for everything else.
How this project evaluates free-to-use AI coding tools: the constraint, the workflow, the evaluation rubric, and what will and will not be claimed.
How far can a zero-cost, AI-assisted development workflow go? This post covers the tools, constraints, and findings from building Pirate Arcade with free models, local hardware, and open source infrastructure.