exam-prep-learning-plan-builder
Pass
Audited by Gen Agent Trust Hub on Aug 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- Indirect Prompt Injection Considerations: The skill processes user-supplied data such as exam topics, syllabus information, and descriptions to build a personalized study plan. This creates a surface where external data enters the agent context. To address this, the instructions include specific guidance for the agent to sanitize the data before embedding it into the HTML template, specifically escaping script termination sequences to prevent potential injection vulnerabilities.
- Secure Data Rendering Practices: The provided HTML template for the study plan application uses secure web development practices. It utilizes
textContentandcreateTextNodeAPIs to render user-controllable strings like topic names and descriptions. These methods ensure that the browser treats the data as literal text, effectively neutralizing the risk of Cross-Site Scripting (XSS) even if the input contains markup-like characters. - Dynamic Content Generation: The skill generates a self-contained HTML application at runtime using Code Interpreter. This process is used to provide a rich user interface (timeline view, calendar picker, and .ics generation) that cannot be achieved through standard chat text. The generation logic is localized to the template provided within the skill assets, following a predictable and safe pattern for asset creation.
- Input Validation and Triage: The skill includes logic to sanity-check user inputs (such as study hours) and provides a 'triage mode' for tight deadlines. These logical checks help ensure the generated output is realistic and useful for the user without introducing security risks.
Audit Metadata