soia-pkm-clip-wechat-article

Pass

Audited by Gen Agent Trust Hub on Jul 20, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes a local Python script scripts/archive_wechat.py. The script follows security best practices by using resolve_contained_path to validate that all file operations are confined within the specified Obsidian vault, preventing path traversal vulnerabilities.- [EXTERNAL_DOWNLOADS]: The skill fetches article data from the web. Network access is strictly controlled by a validation function that requires URLs to use HTTPS and the mp.weixin.qq.com hostname, preventing requests to unauthorized or malicious external services.- [DATA_EXFILTRATION]: Analysis confirmed no data exfiltration risks. The script only accesses the provided URL and the local vault directory. It does not access sensitive system files or credentials.- [PROMPT_INJECTION]: The skill handles untrusted data from the internet, which introduces an indirect prompt injection surface. The risk is mitigated by sanitizing metadata through yaml_escape and clearly delimiting the external content within the generated Markdown files.
  • Ingestion points: External HTML content fetched from URLs in scripts/archive_wechat.py.
  • Boundary markers: Content is placed under the ## 原文 header in the output file.
  • Capability inventory: Network access is restricted to WeChat domains, and file writes are restricted to the vault.
  • Sanitization: Article metadata is escaped for YAML compatibility, and the body is processed using a standard HTML parser.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 20, 2026, 09:18 AM
Security Audit — agent-trust-hub — soia-pkm-clip-wechat-article