wechat-creator
Pass
Audited by Gen Agent Trust Hub on Sep 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill requires the execution of shell commands to install system dependencies (Playwright Chromium) and to run Python scripts for data retrieval, analysis, and report generation.
- [EXTERNAL_DOWNLOADS]: The skill downloads and installs the Chromium browser via Playwright to support the extraction of content from dynamic web pages.
- [INDIRECT_PROMPT_INJECTION]:
- Ingestion points: The skill ingests untrusted data from user-provided URLs and uploaded files (DOCX, TXT, and images via OCR) using
scripts/check_sensitive_words.py. - Boundary markers: Instructions in
SKILL.mdrequire the agent to output raw data in a table format before beginning analysis, providing a level of visibility into the processed content. - Capability inventory: The skill has the capability to perform network requests to
redfox.hkand write diagnostic reports (HTML and TXT) to the local filesystem. - Sanitization: Web content is processed using BeautifulSoup to strip HTML tags, but there is no specific sanitization for natural language instructions that might be embedded in the extracted text.
Audit Metadata