Giving Your Agent Eyes with Game Boy Hacking

2026-07-03 · source · source ↗ · raw capture · wiki page

Ian Langworth describes growing up with an original Game Boy and later a Game Boy Color in an era where consoles were entirely closed systems. Borrowing a game from a friend meant getting "the cart, never the manual." A formative memory: hitting an unbeatable section of a game and only clearing it after stumbling onto a Nintendo Power issue in a random store that happened to cover that exact part. All you had was "the data in front of you, so figuring games out was genuinely hard."

The driving question from childhood: "are there scenes, endings, or content locked away in the ROM that I was never able to reach?" That question, he says, is "exactly the shape of goal you can hand to an agent and let it grind on."

Three Tools

  1. Gearboy — A Game Boy / Game Boy Color emulator built on imgui that exposes everything at runtime: "disassembly, memory views, processor state, sprite sheets, breakpoints, plus the actual playable game."
  2. GhidraBoy — A Game Boy disassembly toolkit for Ghidra.
  3. GhidrAssistMCP — An MCP server in front of Ghidra "so an agent can drive it."

Wired together, Claude can disassemble, investigate, and hunt for exploits in old cartridges. The Game Boy's Sharp LR35902 assembly is noted as simple compared to modern ARM or x86, making it easy for LLMs to reason about.

Working with Claude

Claude demonstrated a solid ability to understand subroutines by "inspecting memory, taking screenshots, and comparing those screenshots over time." Finding straightforward cheats was inconsistent, but that wasn't the goal.

The workflow became a collaborative loop: Claude would set a breakpoint, instruct Ian to play a specific stretch of the game, then ask him to modify a byte and report what changed. Together they mapped out health values for the player's units, the enemy roster and their health, and memory flags checked to determine which screen should display.

The Core Principle

"the thing that matters is the feedback loop." Whether it's a headless Chrome browser or an emulator with a debugger attached, the key is letting an agent observe whether it's approaching its goal. "Give an agent a way to see whether it's achieving its goal, then let it spin. That's when it starts doing surprising things."