The Agent Skills Directory

[SAFE]: The skill implements a strong human-in-the-loop security pattern by explicitly requiring user approval before applying any proposed code modifications.
[SAFE]: The scope of the skill is limited to files modified within the current session, which are identified using standard local tools like git status and git diff.
[SAFE]: References to external guidelines (such as Vercel React best practices) align with industry-standard development workflows and do not involve untrusted remote code execution.
[SAFE]: No indicators of data exfiltration, credential harvesting, or obfuscation were found in the instructions or metadata.

cleanup-web