The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The script scripts/publish.sh contains logic to automatically install the @wenyan-md/cli package from the npm registry if it is not detected on the host system.
[COMMAND_EXECUTION]: The skill executes shell scripts (publish.sh, setup.sh) that trigger external binaries including npm and the wenyan CLI tool.
[CREDENTIALS_UNSAFE]: The skill automatically retrieves sensitive WeChat API credentials (WECHAT_APP_ID, WECHAT_APP_SECRET) from the TOOLS.md file in the user's workspace using grep and sed.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied Markdown content which is then parsed by an external tool. * Ingestion points: The wenyan publish command in scripts/publish.sh takes a user-specified file path as an argument. * Boundary markers: No delimiters or warnings are used to isolate potentially malicious instructions within the Markdown content. * Capability inventory: The skill has access to the local filesystem for reading credentials and Markdown files, and performs network operations to the WeChat API. * Sanitization: Relies entirely on the external @wenyan-md/cli tool for content sanitization.

wechat-publisher