game-designer

Pass

Audited by Gen Agent Trust Hub on Nov 9, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute a local script to capture screenshots of the game for visual analysis. While this is part of the intended audit process, it involves subprocess execution in the developer's environment.\n
  • Evidence: Instruction to run node scripts/iterate-client.js --url http://localhost:3000 --iterations 3 --pause-ms 2000 in SKILL.md.\n- [INDIRECT_PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests untrusted data from the game project and has the capability to modify the project and execute commands. Malicious instructions embedded in the game's source code or metadata could influence the agent's behavior during the audit or improvement phases.\n
  • Ingestion points: Reads package.json, Constants.js, scene files, entity files, and PNG screenshots from the project directory.\n
  • Boundary markers: No delimiters or explicit 'ignore embedded instructions' warnings are specified for the processed game content.\n
  • Capability inventory: Modifies existing code files, creates new files in systems/, entities/, and ui/ directories, and executes subprocesses via node.\n
  • Sanitization: No explicit sanitization, validation, or escaping of the ingested code or file content is performed before the agent processes it.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 9, 2026, 08:17 PM