The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides explicit instructions for the agent to execute shell commands within the user's environment. These include commands for local development testing (mkdocs serve), production deployment to GitHub Pages (mkdocs gh-deploy), and file system operations (unzip, mv, open). These capabilities are essential for the skill's stated purpose but grant the agent significant control over the repository.
[EXTERNAL_DOWNLOADS]: The simulation templates generated by the skill download the p5.js library from well-known and established CDNs including cdn.jsdelivr.net and cdnjs.cloudflare.com. These are legitimate sources for the required framework.
[INDIRECT_PROMPT_INJECTION]: The skill functions as a generator that ingests user-provided educational metadata (objectives, descriptions, lesson plans) and interpolates this content into generated index.md, metadata.json, and main.html files. This creates a surface where malicious instructions embedded in the input data could be written into the project files.
Ingestion points: User-provided inputs for Subject Area, Topic, Learning Objectives, and Descriptions specified in Step 1.
Boundary markers: The skill does not define explicit delimiters or "ignore instructions" markers for the user-supplied content used in file generation.
Capability inventory: The skill has the ability to write multiple file types (.md, .json, .html, .js) to the local file system and execute deployment commands.
Sanitization: The skill automatically converts the MICROSIM_NAME to kebab-case, which provides some protection for file paths, but does not provide specific sanitization for the textual content injected into file bodies.

microsim-p5