review-game

Pass

Audited by Gen Agent Trust Hub on Oct 23, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted external data in the form of game source code, package.json files, and project assets from a path specified in $ARGUMENTS.
  • Ingestion points: Reads game engine configurations, dependency files (package.json), and the main entry points of the codebase.
  • Boundary markers: The instructions do not define specific delimiters or "ignore instructions" wrappers for the code content being analyzed.
  • Capability inventory: The skill performs read-only analysis and generates a structured textual report. It does not possess write or network capabilities itself.
  • Sanitization: There is no evidence of sanitization or filtering for comments or strings within the analyzed code that might contain malicious instructions.
  • [COMMAND_EXECUTION]: In the 'Next Step' section, the skill instructs the agent to suggest a specific shell command to the user: npm run build && ~/.agents/skills/here-now/scripts/publish.sh dist/. While the agent does not execute this command directly, it encourages the user to run a local script from a different skill's directory (here-now), which creates a dependency on the integrity of that external component.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 23, 2026, 12:31 PM