Skills
skills/wwwzhouhui/skills_collection/github-trending/Gen Agent Trust Hub

github-trending

Fail

Audited by Gen Agent Trust Hub on Feb 23, 2026

Risk Level: HIGHCREDENTIALS_UNSAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [CREDENTIALS_UNSAFE]: File send_wecom_summary.py contains a hardcoded sensitive API key in the DEFAULT_WEBHOOK variable (key=45548262-1f73-40a9-a33f-eba95e934082). This allows unauthorized users to send messages through the bot.
  • [DATA_EXFILTRATION]: The skill transmits repository summaries and metadata to the external domain qyapi.weixin.qq.com via a hardcoded webhook URL.
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface through the processing of untrusted repository data.
  • Ingestion points: Fetches raw README content from external, third-party GitHub repositories in fetch_trending.py (via fetch_readme function).
  • Boundary markers: Absent. The skill does not utilize delimiters or specific instructions to treat external README content as data rather than instructions.
  • Capability inventory: Has the ability to perform network operations (sending notifications to an external webhook).
  • Sanitization: The strip_markdown function in send_wecom_summary.py only removes markdown formatting characters and does not sanitize for malicious instructions or hidden control characters.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 23, 2026, 01:07 PM