openmaic-classroom
Pass
Audited by Gen Agent Trust Hub on Mar 17, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill directs the agent to clone the OpenMAIC source code from a third-party GitHub repository (https://github.com/THU-MAIC/OpenMAIC.git) and install dependencies using pnpm.
- [COMMAND_EXECUTION]: Instructions involve running host-level commands including git clone, pnpm install, pnpm dev, and docker compose, which requires significant environment permissions.
- [PROMPT_INJECTION]: The platform is vulnerable to indirect prompt injection during lesson generation. 1. Ingestion points: Topic strings and markdown documents provided via the /api/generate endpoint. 2. Boundary markers: No delimiters are specified to separate user data from instructions. 3. Capability inventory: The system uses dangerouslySetInnerHTML to render custom scenes and generates interactive HTML simulations. 4. Sanitization: No sanitization or HTML escaping for AI-generated scene content is described.
Audit Metadata