teach
Pass
Audited by Gen Agent Trust Hub on Jul 25, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to run CLI commands to open generated HTML lesson files for the user. While intended as a convenience for the user, this provides a surface for shell command execution based on the environment and file paths managed by the skill.
- [PROMPT_INJECTION]: The skill processes multiple user-controlled files (
MISSION.md,RESOURCES.md,learning-records/*.md) to determine its teaching strategy and content, creating a vulnerability to indirect prompt injection. - Ingestion points:
MISSION.md,RESOURCES.md,learning-records/*.md, andNOTES.md. - Boundary markers: None identified in the provided instructions or templates.
- Capability inventory: The agent can write files and execute shell commands to open them.
- Sanitization: No evidence of content validation or sanitization before incorporating file data into the agent's context.
Audit Metadata