xyq-nest-skill

Pass

Audited by Gen Agent Trust Hub on Apr 15, 2026

Risk Level: SAFE
Full Analysis
  • [DATA_EXFILTRATION]: The skill sends user instructions and media files to the official API at xyq.jianying.com for processing. This is a legitimate functional requirement and targets a well-known service domain associated with ByteDance's creative tools.\n- [EXTERNAL_DOWNLOADS]: The download_results.py script downloads generated media from URLs provided by the service's API. These downloads are strictly for user-requested creative assets from verified platform infrastructure.\n- [CREDENTIALS_UNSAFE]: Secrets are managed correctly using the XYQ_ACCESS_KEY environment variable. The skill does not hardcode credentials or implement insecure storage mechanisms.\n- [PROMPT_INJECTION]: Instructions in SKILL.md guide the agent to prioritize this tool for creative tasks. This is standard behavioral configuration for skill routing and does not attempt to bypass security filters or safety guidelines.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes structured data from external API responses, representing a standard interaction surface for integrated tools.\n
  • Ingestion points: Data enters via API responses in scripts/get_thread.py.\n
  • Boundary markers: Not explicitly defined in instructions, but the agent is directed to act as a pass-through for results.\n
  • Capability inventory: The skill has network communication and file system write access for its primary tasks.\n
  • Sanitization: The skill utilizes standard JSON parsing for processing structured API output.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 15, 2026, 12:24 PM