The Agent Skills Directory

[SAFE]: The skill provides templates and localized automation for building interactive engineering handbooks. A detailed review of the included scripts in the assets/templates/scaffold/scripts/ directory confirms they only perform static analysis of the local project files (e.g., regex checks for headers, file existence checks, and date validation).
[COMMAND_EXECUTION]: The utility script eval/run_eval.py utilizes the Python subprocess.run function to execute a local shell script (dry-run-replay.sh) for testing the skill's logic. This is restricted to the development/evaluation environment and does not involve remote code execution or privilege escalation.
[INDIRECT_PROMPT_INJECTION]: The skill identifies and processes user-authored documentation. It includes a security-positive gate (check-no-artifacts.mjs) specifically designed to prevent the leakage of agent tool-call artifacts (such as <parameter> or </content>) into the rendered site, mitigating a common class of indirect injection and content hygiene issues.
[EXTERNAL_DOWNLOADS]: The project configuration specifies standard, well-known dependencies from the official npm registry (e.g., Astro, React, and Starlight) which are standard for modern web development workflows.

starlight-handbook-kit