The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions direct the agent to execute the open command to display generated HTML files in the user's default web browser. Additionally, the diff-review.md template suggests using the git show command to retrieve commit metadata and diff content.
[SAFE]: The skill mandates a 'no external dependencies' rule for its generated HTML files, requiring all CSS and JavaScript to be inlined. This design choice prevents the risk of unauthorized remote code execution or data exfiltration via external script loading.
[SAFE]: The interactive playgrounds are designed to generate natural language prompts that require a human-in-the-loop (the user) to copy and paste them back to the agent, providing a manual validation step before any suggested changes are processed.

playground