decisive-action
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to minimize clarifying questions and proceed autonomously when intent is perceived as clear or standard. This establishes a surface where a malicious actor could craft a request designed to look like a standard operation to trigger the 'Proceed' logic and avoid detection through clarifying questions.
- Ingestion points: The agent processes natural language user requests to evaluate the 'Proceed Threshold Checklist' (SKILL.md).
- Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present to protect the agent from acting on malicious sub-instructions within a request.
- Capability inventory: The skill specifically references destructive operations, file deletion, and data migration as capabilities that might be triggered.
- Sanitization: The guidance lacks instructions to sanitize or validate the user-provided context before the agent determines that intent is 'clear' and proceeds without asking.
Audit Metadata