teach
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill relies on reading content from multiple user-controlled files (MISSION.md, RESOURCES.md, and sequential learning records) to determine its teaching strategy and track user progress. If these files contain instructions designed to influence the agent's behavior, they could be processed as authoritative context.
- Ingestion points: MISSION.md, RESOURCES.md, learning-records/*.md, and GLOSSARY.md are read to establish context and the 'zone of proximal development'.
- Capability inventory: The agent can create and write to markdown and HTML files across the workspace and is encouraged to execute CLI commands to open files.
- Boundary markers: The templates do not define specific delimiters or instructions to ignore embedded commands within the ingested files.
- Sanitization: No explicit sanitization or validation of the content read from these files is mentioned.
- [COMMAND_EXECUTION]: The instructions explicitly direct the agent to execute a CLI command to open generated files for the user.
- Evidence: SKILL.md contains the instruction: 'If possible, open the lesson file for the user by running a CLI command.'
Audit Metadata