baoyu-post-to-wechat

Pass

Audited by Gen Agent Trust Hub on May 18, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes platform-specific commands to facilitate browser automation and system interaction, including osascript on macOS for accessibility and keystroke simulation, powershell.exe on Windows for clipboard and input management, and xdotool or ydotool on Linux.
  • [COMMAND_EXECUTION]: During environment checks and clipboard operations on macOS, the skill dynamically generates and executes local Swift scripts to handle rich media data like images, which is a standard procedure for this platform's clipboard integration.
  • [EXTERNAL_DOWNLOADS]: The article publishing scripts are capable of fetching remote images via HTTP/S if URLs are provided in the article body or metadata, allowing the agent to process and upload these assets to WeChat's servers.
  • [COMMAND_EXECUTION]: The skill uses agent-browser, a specialized CLI tool, to perform high-level browser automation tasks such as navigating the WeChat dashboard and interacting with the editor UI.
  • [DATA_EXFILTRATION]: A feature in the browser-based posting workflow allows users to send the WeChat login QR code to a private Telegram bot if specific bot credentials are provided in the environment; this is a documented convenience feature for remote login scenarios.
Audit Metadata
Risk Level
SAFE
Analyzed
May 18, 2026, 01:22 AM
Security Audit — agent-trust-hub — baoyu-post-to-wechat