iflytek-hyper-tts
Pass
Audited by Gen Agent Trust Hub on Sep 3, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill adheres to secure credential management by using environment variables for authentication, which prevents secret exposure.
- [SAFE]: All network operations are conducted with official vendor domains (*.xf-yun.com), matching the skill's purpose and identity.
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied text as an ingestion surface. The risk is mitigated by its specific function and constraints. 1. Ingestion points: The --text and --text_file arguments in scripts/xfei_hyper_tts.py. 2. Boundary markers: Text is Base64 encoded before being transmitted in a JSON payload via WebSockets. 3. Capability inventory: The script writes to the local file system and connects to external APIs. 4. Sanitization: Includes checks for maximum text size (64KB) and validates that audio parameters are within defined ranges (0-100).
Audit Metadata