wechat-article-aggregator
Pass
Audited by Gen Agent Trust Hub on Feb 26, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Communicates with the external API at
https://down.mptext.topto fetch article metadata and download HTML content. This is the core functionality of the skill as documented. - [PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by ingesting and processing untrusted data from external WeChat articles.
- Ingestion points: Untrusted HTML data is fetched from external URLs via the
download_article_htmlfunction inscripts/fetch_articles.py. - Boundary markers: The resulting output files do not contain explicit delimiters or instructions to prevent an AI from following commands embedded in the text.
- Capability inventory: The skill possesses the ability to perform network requests and write to the local file system, which are necessary for its article collection purpose.
- Sanitization: The tool uses
BeautifulSoupor a dedicatedHTMLParserto target the#js_contentelement, which removes script and style tags, reducing the risk of malicious code execution while maintaining the article's natural language content.
Audit Metadata