doubao-tts

Pass

Audited by Gen Agent Trust Hub on Nov 14, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill acts as a documentation guide for the Doubao TTS service, featuring clear implementation patterns and a dedicated safety section that instructs users and the agent to avoid logging or exposing API credentials. No malicious patterns, obfuscation, or unauthorized access techniques were detected.
  • [EXTERNAL_DOWNLOADS]: The skill correctly references official ByteDance API endpoints at openspeech.bytedance.com for TTS submission and query operations. ByteDance is a well-known technology service provider, and these network operations are documented neutrally as essential functions for the skill's primary purpose.
  • [INDIRECT_PROMPT_INJECTION]: The skill exhibits an ingestion surface for external data via the text parameter in the TTSSelector and DoubaoTTS tool examples. 1. Ingestion points: The text parameter in Python code examples within SKILL.md. 2. Boundary markers: Absent. 3. Capability inventory: Network requests to ByteDance API and local file system writes for audio and metadata. 4. Sanitization: Absent. The risk is assessed as safe because the ingested data is used strictly for audio synthesis and project file output, and the skill does not return this data to the agent's control flow in a way that could influence subsequent logic.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 14, 2026, 08:42 AM