dashscope
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill allows the agent to process external audio files via URL for transcription purposes, creating a potential surface for indirect instructions within the audio content.
- Ingestion points: The
audio_urlparameter inSKILL.mdallows the skill to fetch and process external audio data from any publicly accessible URL. - Boundary markers: The instructions do not define specific delimiters or warnings for the agent to ignore instructions potentially embedded in the resulting transcription data.
- Capability inventory: The skill performs network requests to Alibaba Cloud APIs and writes resulting transcription data to local file paths provided in the
output_pathparameter. - Sanitization: While the skill documentation mentions redacting API keys from error logs, it does not specify sanitization or validation of the transcribed text before it is returned to the agent context.
- [SAFE]: The skill communicates with
dashscope.aliyuncs.com, which is an official endpoint for a well-known cloud service provider. It also correctly advises users to manage sensitive API keys via environment variables rather than hardcoding them.
Audit Metadata