The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to create and write frontend files (index.html, styles.css, script.js) to the workspace root. This is the primary function of the skill. It explicitly forbids writing to sensitive or hidden directories (e.g., node_modules, plugin folders, hidden directories), which follows security best practices for local file system operations.
[DATA_EXFILTRATION]: The skill allows for the ingestion of web search results if available. It contains a specific safety instruction to treat this data as untrusted source material and prevents the external content from overriding the build instructions. This serves as a mitigation against indirect prompt injection.
[PROMPT_INJECTION]: The skill provides specific formatting and parsing instructions for the generated JavaScript (e.g., using JSON-serializable data and escaping characters) to prevent syntax errors or potential script injection within the generated artifact. No instructions were found that attempt to bypass agent safety guidelines.

lesson-generator