The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/check_video.py uses subprocess.run to invoke ffprobe for video metadata analysis. The implementation is secure, using a list of arguments without enabling a shell, which effectively prevents command injection. The tool path is also handled safely.
[EXTERNAL_DOWNLOADS]: The skill's instructions and references target official WeChat domains (channels.weixin.qq.com, res.wx.qq.com) to perform its intended functions. These are well-known and expected service endpoints for a video publishing tool.
[SAFE]: Analysis of the skill's logic, scripts, and instructions revealed no evidence of obfuscation, credential harvesting, or unauthorized data exfiltration. The skill maintains a 'user-in-the-loop' approach for browser authentication and strictly verifies data fields before performing write operations.

lov-publish-wechat-channels