Pinned · October 17, 2026
What a coding agent can actually do in debuff (MCP, live attach, notes, replays, scenarios, fuzz) and what Phaser, PixiJS, KAPLAY, Excalibur and Godot offer agents today.
Read the postPinned · October 17, 2026
A hands-on walkthrough: set up a debuff game with Claude Code, flag a moment and have the agent reproduce it with game_replay, tune params, and verify with scenarios and fuzz.
Read the postA hands-on walkthrough: set up a debuff game with Claude Code, flag a moment and have the agent reproduce it with game_replay, tune params, and verify with scenarios and fuzz.
What a coding agent can actually do in debuff (MCP, live attach, notes, replays, scenarios, fuzz) and what Phaser, PixiJS, KAPLAY, Excalibur and Godot offer agents today.
Build a 2D browser game with Claude Code or another coding agent. The agent can play the game, read its state and check its own changes, and the creative calls stay with you.
keiko runs your real web game in Node with no browser, on a virtual clock, so you can test timing, motion and rules in milliseconds. It works with Phaser, PixiJS, KAPLAY, Excalibur, three.js and debuff.
Write fast, deterministic tests for a Phaser 4 game with keiko: run the real game in Node, hold keys, click, and check movement and timing, with no browser.
How to write game tests that survive tuning: measure time-to-condition with until, check invariants with during, trace motion over time, and sweep params, with keiko.
myaku watches every frame your web game draws, finds the hitches players feel, and explains each one in plain words: your script and where, garbage collection, GPU uploads, shader compiles or browser throttling. One import, any engine.
Why does a three.js scene stutter the first time something appears? Shader compiles, texture uploads and garbage collection, found and explained per frame by myaku, a free frame profiler for web games.
Find out why a Phaser 3 or Phaser 4 game drops frames: which scene's update is slow, texture uploads, shader compiles, garbage collection or the browser. One import with myaku, a free frame profiler for web games.
Find out why a PixiJS 8 game drops frames: ticker updates versus rendering, texture uploads, garbage collection or the browser. myaku hooks into Pixi automatically, with one import.
debuff and Babylon.js compared for browser games: 2D versus 3D, built-in tools like the Playground and Inspector, physics, download size, testing and coding-agent support.
debuff and Excalibur.js compared for 2D browser games: built-in features, size, testing tools and coding-agent support, with where Excalibur is ahead.
debuff and KAPLAY (the successor to Kaboom.js) compared: API style, size, playground, testing and coding-agent support, and when each fits.
debuff and LittleJS compared: features, size, jam friendliness, procedural sound, level editing, testing and coding-agent support.
debuff and melonJS compared for 2D browser games: renderers, Tiled support, physics, size, testing and coding-agent support.
debuff and Phaser compared for 2D web games: features, maturity, editor, bundle size, testing and coding-agent support, with where Phaser is the better pick.
An honest comparison of debuff, PixiJS and Phaser for 2D browser games: what each is, bundle size, rendering speed, tooling and AI agent support.
debuff and PixiJS compared for 2D browser games: what each includes, measured sprite throughput, bundle size, testing and coding-agent support.
debuff and three.js compared for browser games: a 3D rendering library versus a 2D game toolkit, size, game features, testing and coding-agent support.
debuff and Cocos Creator compared: a TypeScript toolkit versus a TypeScript engine with an editor, web and mobile builds, testing and coding-agent support.
debuff and Construct 3 compared for 2D browser games: event sheets versus TypeScript, pricing and free-tier limits, HTML5 export, testing and coding-agent support.
debuff and Defold compared for 2D browser games: Lua and an editor versus TypeScript, HTML5 export, licensing, testing and coding-agent support.
debuff and GameMaker compared for 2D games on the web: GML and an IDE versus TypeScript, HTML5 and GX.games exports, pricing, testing and AI tooling.
debuff and GDevelop compared: code versus no-code events, built-in AI agent versus bring-your-own coding agent, publishing, testing and pricing.
debuff and Godot's web export compared for 2D browser games: editor, features, web export constraints, testing and coding-agent support, and when to use each.
How fantasy consoles PICO-8 and TIC-80 compare with debuff for small browser games: creative constraints, built-in editors, sharing, web export and testing.
debuff and Unity's Web platform compared for 2D browser games: editor and ecosystem, web build constraints, WebGPU status, pricing, testing and AI agent support.
Make a 2D pixel-art platformer that runs in any browser: tilemaps from Tiled or LDtk, crisp pixels, and a jump that feels good thanks to coyote time and jump buffering.
Games are hard to test, but deterministic games aren't. How a fixed-step clock and seeded RNG enable scenario tests, exact replays and fuzz playtesting for 2D web games.
How Flash and portals like Newgrounds, Miniclip, Kongregate and Armor Games shaped browser games, how Flash ended, how its games are preserved, and what HTML5 games carry on.
