miniprogram-development
Pass
Audited by Gen Agent Trust Hub on Nov 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to use official CLI tools including
wechatide(for WeChat Developer Tools interaction) andminiprogram-ci(for CI/CD workflows like uploading and previewing). These tools are standard for the WeChat ecosystem. - [REMOTE_CODE_EXECUTION]: The skill configuration references the use of
npxto execute@cloudbase/npm-launcherand@cloudbase/cloudbase-mcp. These packages are official vendor-owned resources under the@cloudbasescope, which corresponds to the skill author (tencentcloudbase). - [EXTERNAL_DOWNLOADS]: The skill references official documentation and binary download locations at
developers.weixin.qq.comanddocs.cloudbase.net. These are trusted, well-known domains for WeChat and CloudBase development. - [INDIRECT_PROMPT_INJECTION]: The skill processes project-specific files like
project.config.jsonand mini program source code. While this presents an indirect prompt injection surface, it is the primary intended function of a development skill, and the instructions include defensive measures such as explicit user confirmation before performing sensitive operations.
Audit Metadata