start-super-pm
Pass
Audited by Gen Agent Trust Hub on May 17, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: Instructional Override Patterns
- The skill uses authoritative and forceful language designed to override the agent's standard decision-making processes, including terms like "EXTREMELY-IMPORTANT", "MUST", "non-negotiable", and "RED ALERT".
- It explicitly instructs the agent to disregard its own judgment ("This is non-negotiable... you can't find excuses to skip") in favor of the skill's routing logic.
- [PROMPT_INJECTION]: Indirect Prompt Injection Surface
- Ingestion points: The skill monitors all incoming user messages to identify task types (e.g., "I want to do X") as described in the "Task Type Identification Guide" in
SKILL.md. - Boundary markers: Absent. The skill does not employ delimiters or specific instructions to isolate user-provided data from the agent's operational logic when invoking tools.
- Capability inventory: The skill is designed to automatically trigger a wide array of other functional skills using the platform's
Skilltool (e.g.,pm-demand,pm-docs,pm-search), providing a broad functional surface for any injected instructions. - Sanitization: Absent. There is no evidence of input validation or content filtering before user-supplied strings are used to determine which tool to execute and what arguments to pass.
Audit Metadata