The Agent Skills Directory

[DYNAMIC_EXECUTION]: The auditing tool uses scripts/lib/app-modules.mjs to dynamically load and execute application source code using Node's module.registerHooks. This is employed to run the real game reducers and rules against a test suite during analysis.
[DYNAMIC_EXECUTION]: The script scripts/lib/zustand-stand-in.mjs generates a mock module for the zustand state management library at runtime. It creates this module from a hardcoded string literal and loads it using a data: URL to provide a controlled environment for testing without requiring external dependencies.
[INDIRECT_PROMPT_INJECTION]: The skill presents an attack surface where malicious instructions could be embedded in the repository code being analyzed.
Ingestion points: The checker script scripts/check-game-host.mjs loads and reads source files from the target repository via scripts/lib/host-layout.mjs and scripts/lib/app-modules.mjs.
Boundary markers: The auditing logic does not employ explicit delimiters or system instructions to isolate the content of the analyzed files from the analyzer's own instructions.
Capability inventory: The analysis scripts possess the capability to execute code (via app-modules.mjs) and run subprocesses (via spawnSync in scripts/check-lib.mjs).
Sanitization: There is no evidence of filtering or sanitization performed on the ingested code or text content before it is processed by the agent.

game-session-host