The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on executing local Python scripts (scripts/fetch_articles.py and scripts/generate_hot_html.py) via the command line to retrieve data and generate visualization reports.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from an external API (redfox.hk), including article titles and author names. This content is displayed in the agent's response and embedded into generated HTML files without escaping or sanitization, creating a surface for injection.
Ingestion points: scripts/fetch_articles.py retrieves article metadata from a remote API.
Boundary markers: None identified in the output instructions to prevent the agent from interpreting embedded instructions in article titles.
Capability inventory: The skill has capabilities for network access, file writing, and command execution.
Sanitization: The HTML generation script (scripts/generate_hot_html.py) uses f-strings to insert titles and account names directly into HTML templates without proper encoding, potentially allowing for Cross-Site Scripting (XSS) in the generated reports.
[EXTERNAL_DOWNLOADS]: The skill performs network operations to fetch data and external resources.
It connects to https://redfox.hk/story/api/cozeSkill/getWxDataByCategoryAndTime to retrieve WeChat article rankings.
In scripts/fetch_articles.py, the script explicitly disables SSL verification by setting ssl.CERT_NONE and check_hostname = False. This weakens the security of the communication channel and makes the connection susceptible to Man-in-the-Middle (MITM) attacks.
The generated HTML report includes a remote script dependency from a well-known CDN (cdnjs.cloudflare.com/ajax/libs/html2pdf.js/0.10.1/html2pdf.bundle.min.js).

wechat-original-hot