soia-pkm-clip-wechat

Pass

Audited by Gen Agent Trust Hub on Jul 14, 2026

Risk Level: SAFE
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill fetches HTML content from mp.weixin.qq.com. This network activity is strictly controlled by a validator that enforces the HTTPS protocol, the specific WeChat hostname, and standard ports, effectively preventing SSRF or unauthorized external connections.
  • [COMMAND_EXECUTION]: The skill utilizes a local Python script (scripts/archive_wechat.py) to process article data. The script uses only standard libraries and includes internal checks to ensure that all file writes are contained within the authorized vault directory.
  • [DATA_EXFILTRATION]: There is no evidence of data exfiltration. The script is designed for local archiving, and its network usage is limited to retrieving content from a validated source. It includes logic to handle metadata and content locally without transmitting sensitive information to third-party servers.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external HTML content, which constitutes an indirect prompt injection surface. However, it implements robust mitigations: it uses a formal HTML parser to extract text rather than vulnerable regex, strips potentially malicious zero-width characters, and escapes YAML metadata to prevent frontmatter injection. Content is clearly delimited within the final Markdown file using specific headers and markers.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 14, 2026, 03:44 PM
Security Audit — agent-trust-hub — soia-pkm-clip-wechat