The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted Markdown files provided by the user and presents a preview of the converted content back to the agent's context. A malicious document could contain hidden instructions intended to manipulate the agent's subsequent actions.
Ingestion points: Markdown files read via the input argument in scripts/md_to_wechat.py.
Boundary markers: The skill does not use specific delimiters or instructions to warn the agent about potentially malicious content within the processed data.
Capability inventory: The skill environment allows file reading/writing, shell command execution for environment setup, and opening the system browser.
Sanitization: While the skill uses BeautifulSoup for HTML structure, it does not sanitize the text content of the Markdown for potential injection attacks.- [EXTERNAL_DOWNLOADS]: The skill references and fetches resources from external domains associated with the target platform.
Fetches a background icon from the official WeChat CDN (mmbiz.qpic.cn) for styling code blocks in scripts/md_to_wechat.py and scripts/optimize_wechat.py.
Includes a navigational link to the official WeChat Official Account login page (mp.weixin.qq.com) in the generated preview.- [COMMAND_EXECUTION]: The skill instructions require the agent to execute shell commands to manage the execution environment.
Commands include python3 -m venv, pip install -r requirements.txt, and direct execution of the conversion scripts within the virtual environment.

haizei-wechat-article-convert