test
Pass
Audited by Gen Agent Trust Hub on Apr 10, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: The skill uses emphatic language such as 'MANDATORY IMPORTANT MUST' and 'IMPORTANT' to enforce specific agent behaviors like task decomposition and evidence-based reasoning. These instructions are focused on task performance and do not attempt to override safety filters or hijack the underlying model's behavior.
- [DATA_EXFILTRATION]: No network activity or sensitive file access was detected. The skill reads project-specific documentation and writes to a local reports directory, which are standard operations for development tools.
- [COMMAND_EXECUTION]: The skill delegates testing functionality to an encapsulated 'tester' subagent rather than executing direct shell commands, minimizing the risk of arbitrary code execution.
- [SAFE]: The skill follows security best practices by implementing a 'READ-ONLY' rule for its testing phase and requiring user validation for decisions.
Audit Metadata