The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to use shell commands (Bash) to locate the most recent Markdown files in the workspace and execute Python scripts for processing and formatting.
[EXTERNAL_DOWNLOADS]: The skill requires standard Python libraries (markdown, beautifulsoup4, cssutils, lxml, watchdog, Pygments) which are common for text processing and file monitoring.
[DATA_EXFILTRATION]: The preview_generator.py script starts a local HTTP server using Python's built-in HTTPServer. This server is bound to localhost and is used exclusively for the live preview feature of the converted HTML files.
[DATA_EXPOSURE]: The skill reads local Markdown files and templates to perform its conversion tasks. It does not attempt to access sensitive system directories or credentials.
[INDIRECT_PROMPT_INJECTION]: The skill acts as an ingestion point for user-provided Markdown content. While it processes untrusted data and has access to shell tools, its behavior is limited to text transformation (Markdown to HTML) and does not interpret input content as executable instructions for the agent.
Ingestion points: User-supplied Markdown files.
Boundary markers: None explicitly mentioned in the conversion logic, though content is parsed via a standard Markdown library.
Capability inventory: Shell access (Bash), file read/write access (Read, Write).
Sanitization: Content is parsed using the markdown and BeautifulSoup libraries, which transform text into structured HTML without executing embedded logic.

wechat-article-formatter