education-platforms
Pass
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: LOWNO_CODE
Full Analysis
- [No Executable Code] (INFO): The skill package is composed strictly of Markdown reference files and configuration metadata. There are no Python scripts, Node.js packages, or shell commands present, which eliminates the risk of RCE, persistence, or privilege escalation.
- [Indirect Prompt Injection] (LOW):
- Ingestion points: The skill processes user-supplied text when keywords like 'lms' or 'course platform' are detected (defined in
SKILL.md). - Boundary markers: No specific delimiters are used to separate user input from the platform-provided references.
- Capability inventory: No capabilities for file writing, command execution, or network communication are present in the skill.
- Sanitization: No sanitization logic is implemented.
- Assessment: While an injection surface exists, the impact is negligible as the agent has no external capabilities to exploit; the risk is limited to biased reasoning within the session.
Audit Metadata