The Agent Skills Directory

[COMMAND_EXECUTION]: The Python utility script utilizes a subprocess call to execute the system's curl binary. This is used to retrieve the raw HTML content of the WeChat article provided by the user.
Evidence: subprocess.run(["curl", "-s", "-A", user_agent, url], ...) in scripts/extract.py.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest data from external, third-party URLs which are then processed by the AI to generate styling code. This code is automatically written to a local file (markdown-to-wechat.html) in the user's workspace, creating a vulnerability surface where malicious content in an article could attempt to manipulate the file modification process.
Ingestion points: The scripts/extract.py script fetches external content and saves it to .extracted_content.html for AI analysis.
Boundary markers: None identified. The instructions do not specify the use of delimiters or safety warnings for the AI when reading the extracted content.
Capability inventory: The skill allows the agent to read local files and modify the markdown-to-wechat.html file, including appending or updating JavaScript objects and CSS styles.
Sanitization: No sanitization of the fetched HTML content is performed beyond basic regex extraction of specific tags, leaving the content as raw input for the AI's context.

wechat-theme-extractor-cn