The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes system commands using Bun.spawn and Bun.spawnSync. Specifically, it uses open, xdg-open, or explorer to launch the browser for the preview interface, and the sips utility on macOS to compress images that exceed size limits. These actions are restricted to legitimate operational requirements.
[EXTERNAL_DOWNLOADS]: The skill facilitates the migration of remote images to the WeChat ecosystem by downloading external assets via fetch. This behavior is expected given the skill's purpose of synchronizing local Markdown content with the WeChat draft box.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-provided Markdown files which may contain untrusted data. This creates a potential attack surface for indirect prompt injection. However, the skill implements disable-model-invocation: true and focuses on automated processing, significantly reducing the risk of the agent executing malicious instructions embedded in the processed data.
[SAFE]: The skill follows security best practices for credential management by utilizing .env files within the user's local configuration directory instead of hardcoding secrets. It also implements path validation and MIME type sniffing to ensure only validated image files are processed and uploaded.

wechat-publish