The Agent Skills Directory

[SAFE]: The skill incorporates strong safety guardrails, explicitly instructing the AI agent to ask for user permission before installing Python libraries like sympy or rdkit if they are missing from the environment.
[SAFE]: A human-in-the-loop verification process is required when processing chemical reactions from images, ensuring the user confirms identified data before any code execution or file generation occurs.
[SAFE]: The instructions mandate the immediate shutdown of any local web servers or ports opened for previewing the generated HTML, following best practices to prevent leaving unauthorized services running on the host machine.
[EXTERNAL_DOWNLOADS]: The HTML template utilizes trusted CDNs (Cloudflare and jsDelivr) to load essential frontend libraries such as Three.js and KaTeX. These are well-known technology services and are used appropriately for the skill's primary purpose of molecular visualization.
[COMMAND_EXECUTION]: The skill utilizes the local python3 interpreter for legitimate scientific computing tasks, including equation balancing and geometric calculations, which are performed within the scope of the skill's intended functionality.

edu-chem-reaction