exam-answer-eval
Pass
Audited by Gen Agent Trust Hub on Jul 3, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill is designed for pedagogical coaching, guiding the agent to provide constructive feedback on user-provided text. It does not request any sensitive permissions or use dangerous tools.
- [PROMPT_INJECTION]: The skill ingests user-supplied text (exam answers and prompts). This is an inherent part of its functionality. While it lacks explicit boundary markers to delimit user input from instructions, the agent is restricted to educational feedback tasks and does not have access to capabilities that could be abused via injection, such as network access or shell execution.
- [NO_CODE]: The skill consists entirely of instructional Markdown and YAML configuration. It does not contain scripts, binaries, or references to external code execution.
Audit Metadata