using-superpowers
Pass
Audited by Gen Agent Trust Hub on Feb 22, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- Prompt Injection (LOW): The skill employs extremely forceful and authoritative markers ("EXTREMELY-IMPORTANT", "not negotiable", "not optional") typical of instructions designed to override an AI's internal safety and reasoning protocols. It mandates that the agent prioritize tool execution even when there is only a "1% chance" of relevance.\n- Indirect Prompt Injection Surface (LOW): By demanding that skills be invoked before clarifying questions or context gathering, the skill removes critical safety checks and human-in-the-loop validation. This creates a vulnerability where the agent might blindly execute instructions from untrusted external files before identifying potential risks.\n
- Ingestion points: Any user message or file content identified as a potential "skill".\n
- Boundary markers: Absent; the skill explicitly instructs the agent to ignore standard reasoning steps like seeking more context first.\n
- Capability inventory: The Skill tool, which interprets and follows content from external skill files.\n
- Sanitization: None; the agent is instructed to follow skills "exactly" without adapting away discipline.
Audit Metadata