67
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill uses standard reasoning patterns to simulate a multi-agent environment. No hardcoded credentials, malicious network operations, or obfuscated payloads were detected.
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-defined task briefs and fans them out to multiple sub-agent personas, which is a potential surface for instructions within the task data to influence sub-agent behavior.
- Ingestion points: The 'task brief' defined in the 'Frame once' behavior section.
- Boundary markers: None; the brief is passed verbatim to sub-agents.
- Capability inventory: No file system access, network requests, or shell command execution detected in the skill logic.
- Sanitization: No explicit sanitization or escaping of the user-provided brief is performed before interpolation into sub-agent prompts.
- [COMMAND_EXECUTION]: The skill mentions an optional 'Agent/Task tool' and provides role-play instructions for sequential passes. These are internal agent orchestration patterns and do not involve arbitrary shell execution.
Audit Metadata