ask-questions-if-underspecified
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The instructions mandate a 'pause before acting' phase, which prevents the agent from running commands or editing files while must-have unknowns remain. This is a positive safety pattern.
- [DATA_EXPOSURE]: The skill references local project configuration files like REPO_PROFILE.json and AGENTS/CALIBRATION for context, which is standard behavior for project-aware agents and does not involve sensitive system-level credentials.
- [INDIRECT_PROMPT_INJECTION]: While the skill processes external user input, its primary function is to serve as a guardrail against premature execution. The instructions specifically guide the agent to restate requirements and await confirmation, reducing the risk of malicious instructions in user data being executed without oversight.
Audit Metadata