The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the installation of components from an external GitHub repository (github.com/WaterTian/wechat-devtools-mcp) and a package from a public registry via uv.
[COMMAND_EXECUTION]: The skill interacts with the local file system and operating system by executing commands through the WeChat Developer Tools CLI and process management tools like taskkill.
[DYNAMIC_EXECUTION]: The wechat_automator tool facilitates the execution of arbitrary JavaScript strings using the evaluate action and allows modification of API behaviors via the mock_wx action within the WeChat IDE.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data such as mini-program source code and runtime console logs, which could serve as a vector for indirect prompt injection.
Ingestion points: wechat_file (reads WXML, WXSS, JS, and JSON project files) and wechat_inspector (captures console and CDP logs).
Boundary markers: The instructions do not define specific delimiters or warnings to ignore instructions embedded within the source files or logs being analyzed.
Capability inventory: The skill possesses high-impact capabilities including wechat_build (uploading code to production), wechat_automator (arbitrary JS execution), and wechat_file (reading local project structure).
Sanitization: There is no evidence of sanitization or filtering of the content retrieved from project files or runtime logs before it is processed by the agent.

wechat-devtools-mcp