wechat-message
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPRIVILEGE_ESCALATIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The scripts
scripts/wechat_automation_script.applescriptandscripts/wechat_automation_script.ps1download image files from user-supplied URLs usingcurlandInvoke-WebRequestrespectively. - [COMMAND_EXECUTION]: The macOS script utilizes
do shell scriptto execute various system commands includingcurlfor downloads,mktempfor temporary directory creation,rmfor cleanup, andsipsfor image processing. It also relies on the externalcliclickutility for mouse automation. - [DYNAMIC_EXECUTION]: The Windows PowerShell script uses
Add-Typeto dynamically compile C# code for P/Invoke, enabling access to Win32 API functions inuser32.dllfor window management and mouse events. - [PRIVILEGE_ESCALATION]: The skill requires the user to manually grant 'Accessibility' permissions to their terminal or application environment in macOS System Settings. This permission allows the script's
System Eventscommands to interact with and control the WeChat user interface. - [INDIRECT_PROMPT_INJECTION]: The skill accepts message content and contact names from the agent's context. While it uses proper quoting for shell commands, if the agent receives instructions from an untrusted external source to send a message, the skill will execute that automation as requested. Boundary markers are not explicitly defined for the message payload.
Audit Metadata