using-superpowers
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFEPROMPT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill employs extremely forceful and mandatory language, such as 'ABSOLUTELY MUST', 'NOT NEGOTIABLE', and 'YOU DO NOT HAVE A CHOICE', to compel the agent to prioritize the 'Superpowers' workflow. These instructions are specifically designed to override the agent's default reasoning and autonomy regarding tool invocation, requiring it to check for and load skills before performing any other task or asking clarifying questions.
- [INDIRECT_PROMPT_INJECTION]: This skill functions as a foundational orchestration layer that mandates the ingestion of external data from other skill files using platform-specific tools like
Skilloractivate_skill. This creates an inherent vulnerability surface where the agent could be influenced by instructions contained within those activated skill files. - Ingestion points: Data is loaded into the agent's context from external markdown files via the platform's skill-loading mechanisms.
- Boundary markers: The skill does not define or enforce any delimiters or security boundaries for the content it instructs the agent to load.
- Capability inventory: While this specific skill does not directly execute dangerous commands, it establishes the mechanism for the agent to load and follow instructions that may possess those capabilities (e.g., file system access, command execution).
- Sanitization: There are no instructions for validating, filtering, or sanitizing the content of the skills being activated.
Audit Metadata