teach
Pass
Audited by Gen Agent Trust Hub on Jul 3, 2026
Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection. It is designed to read and process external resources (URLs) provided by the user or found online to generate educational content. Malicious instructions embedded in these external sources could influence the agent's behavior during lesson creation. \n
- Ingestion points: RESOURCES.md (external URLs), MISSION.md. \n
- Boundary markers: No explicit markers or "ignore instructions" directives are used when processing external content. \n
- Capability inventory: File writing (./lessons/, ./assets/, NOTES.md), CLI command execution (opening files). \n
- Sanitization: No sanitization or validation of the external content is prescribed before it is incorporated into the workspace.\n- [COMMAND_EXECUTION]: The instructions direct the agent to generate interactive HTML lessons and reusable components, such as JavaScript-based quizzes and simulators. This involves the dynamic generation of executable code for the user's environment.\n- [COMMAND_EXECUTION]: The skill instructs the agent to run CLI commands to open lesson files for the user, which is a standard but noteworthy capability.
Audit Metadata