The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's workflow involves fetching external information to generate interactive content, which establishes a surface for indirect prompt injection. 1. Ingestion points: External URLs and primary sources referenced in RESOURCES.md and SKILL.md. 2. Boundary markers: The skill relies on source selection but does not specify technical boundary markers to isolate external data from the agent's instructions. 3. Capability inventory: Writing HTML lessons, Markdown records, and assets to the filesystem, plus the instruction to open files using CLI commands. 4. Sanitization: No explicit sanitization or filtering of external content is defined for the lesson generation process.
[DYNAMIC_EXECUTION]: The skill instructs the agent to create interactive components like "simulators" and "quiz widgets" in HTML/JS format within the ./assets/ and ./lessons/ directories. This involves the generation and subsequent rendering of script content.
[COMMAND_EXECUTION]: The agent is prompted to use a CLI command to open generated lesson files for the user. While intended for user convenience, this capability provides a command execution vector in the local environment.

teach