hinge-question-designer
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from several external fields and interpolates them directly into the instructions provided to the agent.
- Ingestion points: The skill uses
concept_being_taught,student_level,lesson_context,known_misconceptions,student_profiles,response_method, andtime_constraintas input variables inSKILL.md. - Boundary markers: The prompt utilizes markdown bolding and headers (e.g.,
**Concept:** {{concept_being_taught}}) to separate the input data from the instructions, but it lacks explicit 'delimiter' warnings or instructions to ignore any commands potentially embedded within the student or lesson data. - Capability inventory: The skill is restricted to text generation; it does not request tool permissions, network access, or file-writing capabilities in its frontmatter or prompt body.
- Sanitization: There is no evidence of validation or sanitization of the input strings before they are processed by the model.
Audit Metadata