The Agent Skills Directory

[COMMAND_EXECUTION]: The agent is instructed to use CLI commands to open generated HTML lesson files for the user. This instruction could be exploited to execute arbitrary shell commands if the agent is manipulated into using a dangerous utility or malformed file path.
[REMOTE_CODE_EXECUTION]: The skill involves the dynamic generation of interactive HTML lessons and reusable components (such as simulators and quiz widgets) containing JavaScript. This runtime creation of executable code poses a risk if the generation process is influenced by malicious instructions.
[PROMPT_INJECTION]: The skill defines a workflow for ingesting and processing data from external, untrusted sources (URLs) to ground its teaching content, creating a surface for indirect prompt injection.
Ingestion points: External websites and documents linked in RESOURCES.md and searched during knowledge acquisition.
Boundary markers: There are no explicit instructions for using delimiters or warnings to ignore instructions within the retrieved external content.
Capability inventory: The agent has the ability to write multiple file types (Markdown, HTML, JS) to the local filesystem and execute CLI commands.
Sanitization: The instructions lack requirements for sanitizing or escaping external content before it is interpolated into the generated lessons or internal records.

teach