security-audit
Pass
Audited by Gen Agent Trust Hub on Aug 24, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill uses natural language instructions like 'Never assume a pattern is safe' and 'CRITICAL security findings must be resolved,' which mimic prompt injection control patterns. While intended for safety, these could be manipulated if the agent interprets them as overrides to its own core behavioral constraints.
- [COMMAND_EXECUTION]: The skill utilizes the 'Bash' and 'Task' tools to perform system-wide scans and grep operations. It explicitly instructs the agent to execute shell commands based on search patterns, which is a powerful capability that could be abused if combined with malicious input.
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process a wide variety of untrusted project files including source code (
src/), configuration files, and asset data (assets/data/). It lacks explicit boundary markers or sanitization instructions for how the agent should handle potentially malicious instructions embedded within the audited game code or metadata.
Audit Metadata