ad-copywriting
Pass
Audited by Gen Agent Trust Hub on Mar 25, 2026
Risk Level: SAFE
Full Analysis
- [DATA_EXFILTRATION]: Performs network requests to Volcengine's official API (ark.cn-beijing.volces.com) to process content generation tasks. This communication is essential for the skill's primary function and targets a well-known service provider.
- [PROMPT_INJECTION]: The skill processes untrusted product data from user input and interpolates it into LLM prompts. The absence of boundary markers or input validation in scripts/generate_copy.py creates a surface for indirect prompt injection via provided product descriptions.
- [COMMAND_EXECUTION]: Implements file writing operations to save generated advertisement variants and task events to the local filesystem based on user-provided path arguments.
Audit Metadata