The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill uses several techniques to execute code dynamically during the development and simulation phases.
scripts/econ-sim.mjs uses the new Function constructor to evaluate and run the game's core logic (00-core.js and 50-sim.js) for bot simulation.
scripts/shot.mjs and scripts/preview.mjs generate temporary HTML files containing the skill's source code and execute them within a headless Chrome environment.
scripts/shot.mjs reads a developer-provided script path from command-line arguments and injects its content into the execution context.
[INDIRECT_PROMPT_INJECTION]: The skill's workflow involves analyzing external reference materials, which creates a vulnerability surface for indirect instructions.
Ingestion points: Instructions in SKILL.md direct the agent to study external game URLs, source folders, and video frames to extract design mechanics.
Boundary markers: Absent; no specific instructions are provided to the agent to disregard or sanitize instructions embedded in reference documentation or code.
Capability inventory: The skill has extensive capabilities, including file system access (readFileSync, writeFileSync), shell command execution via execFileSync, and dynamic code evaluation.
Sanitization: Absent; the skill processes external reference data without evident filtering or validation steps.
[COMMAND_EXECUTION]: The utility scripts (shot.mjs and preview.mjs) use execFileSync to launch a headless browser (Google Chrome) and ffmpeg to process video frames. While these are necessary for the skill's primary function of game prototyping, they represent a high-privilege command execution surface.
[EXTERNAL_DOWNLOADS]: The skill's HTML template (assets/starter/src/page.html) references external font assets from Google Fonts (fonts.googleapis.com and fonts.gstatic.com). These are well-known and trusted services used for UI rendering.

game-prototype-bible