The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions explicitly direct the agent to 'open the lesson file for the user by running a CLI command' after creating it. This introduces a capability for shell command execution within the workspace.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to gather knowledge from external 'high-quality, high-trust resources' and use that information to generate lessons. This creates a vulnerability surface where malicious content on a linked website could attempt to influence the agent's output or behavior.
Ingestion points: External URLs documented in RESOURCES.md and content fetched from the web.
Boundary markers: None identified in the instructions to delimit external content from system instructions.
Capability inventory: File system writes (creating markdown and HTML files) and CLI command execution (opening files).
Sanitization: There are no instructions for sanitizing, escaping, or validating the content retrieved from external sources before it is processed into lessons.
[DYNAMIC_EXECUTION]: The skill involves the generation of interactive HTML components, including 'quiz widgets, simulators, and diagram helpers' stored in an ./assets/ directory. These generated assets likely contain JavaScript that is executed in the user's browser context when the lessons are opened.

teach