The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest information from external sources (as defined in RESOURCES.md) and use that information to author HTML lessons. This creates a vulnerability surface where malicious instructions or scripts embedded in those external sources could be included in the generated lessons.
Ingestion points: External resources and websites referenced in RESOURCES.md and MISSION.md.
Boundary markers: None identified in the instructions for separating external content from agent-authored instructions.
Capability inventory: Writing HTML files to ./lessons/ and creating reusable components (CSS/JS) in ./assets/ (SKILL.md).
Sanitization: There are no instructions to sanitize, filter, or escape data retrieved from external sources before it is interpolated into the generated HTML files.
[DYNAMIC_EXECUTION]: The skill generates executable web content, including HTML lessons and reusable assets like scripts or simulators, which are then stored locally for user interaction.
Evidence: The instructions in SKILL.md direct the agent to create "self-contained HTML output" in ./lessons/ and write "reusable components" such as "stylesheets, quiz widgets, simulators, diagram helpers" in ./assets/.
[COMMAND_EXECUTION]: The skill instructs the agent to open generated lesson files using system CLI commands.
Evidence: SKILL.md contains the instruction: "If possible, open the lesson file for the user by running a CLI command."

teach