using-superpowers
Pass
Audited by Gen Agent Trust Hub on Mar 12, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill uses extremely strong imperative language (e.g., 'ABSOLUTELY MUST', 'YOU DO NOT HAVE A CHOICE', 'not negotiable', 'not optional') to override the agent's default reasoning and force the invocation of tools.
- [PROMPT_INJECTION]: Establishes a '1% rule' that commands the agent to invoke skills even if there is only a marginal chance they apply. This behavioral override is designed to bypass the agent's standard judgment on whether a tool is necessary.
- [SAFE]: The skill explicitly defines an 'Instruction Priority' where user instructions (like CLAUDE.md or direct requests) always take precedence over the skill's own rules, which acts as a safety mechanism against total agent hijacking.
- [SAFE]: No hardcoded credentials, network exfiltration, or unauthorized file system access patterns were found in the instructions or tool mappings.
Audit Metadata