publish-zsxq-article

Pass

Audited by Gen Agent Trust Hub on May 13, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses local shell commands (cat, ls) and a Python script to manage local files (reading Markdown, checking image sizes). These are standard operations for the skill's stated purpose of publishing local content.
  • [REMOTE_CODE_EXECUTION]: The skill uses evaluate_script (Chrome DevTools) or browser_console_exec (Playwright) to interact with the Zsxq website. This is used solely to simulate a 'paste' event in the Milkdown editor to ensure proper Markdown rendering, which is a legitimate technical requirement of the platform's editor.
  • [SAFE]: All network operations are directed to the official Zsxq domain (wx.zsxq.com) for article publishing. There is no evidence of data exfiltration or unauthorized access. The skill explicitly instructs the agent to save as a draft and never auto-publish, maintaining user control.
Audit Metadata
Risk Level
SAFE
Analyzed
May 13, 2026, 02:10 PM