using-supaconductor
Pass
Audited by Gen Agent Trust Hub on Apr 6, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill employs extremely forceful language such as 'ABSOLUTELY MUST', 'YOU DO NOT HAVE A CHOICE', and 'This is not negotiable' to override the agent's standard decision-making regarding tool selection and task prioritization.
- [PROMPT_INJECTION]: It explicitly instructs the agent to ignore its own 'rationalizations' or doubts, which is a pattern commonly used to bypass logical or safety-oriented reasoning filters.
- [PROMPT_INJECTION]: The instructions mandate that the agent invoke the 'Skill' tool before any other response or action, including asking the user for clarification, which reduces user oversight and suppresses the agent's discretion.
- [PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection by forcing the agent to load and follow any skill with even a '1% chance' of applying. Ingestion points: External skill files (SKILL.md) loaded via the 'Skill' tool. Boundary markers: None; the agent is told to 'follow skill content directly'. Capability inventory: Facilitates the invocation of other skills that may have capabilities like file access or subprocess execution. Sanitization: None mentioned; the skill does not validate or sanitize the content of loaded skills before execution.
Audit Metadata