aliyun-qwen-asr
Pass
Audited by Gen Agent Trust Hub on Apr 29, 2026
Risk Level: SAFEPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- [DATA_EXFILTRATION]: The skill accesses sensitive local files such as
.envand~/.alibabacloud/credentialsto obtain API keys for authentication. These credentials are submitted to official Alibaba Cloud (Aliyun) DashScope endpoints to perform transcription tasks, which is consistent with the skill's intended functionality as a service provider. - [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by processing untrusted audio data from external sources.
- Ingestion points: Audio input provided via URL or file path to the Python helper script (scripts/transcribe_audio.py) and subsequent transcription JSON fetched from external links.
- Boundary markers: The skill does not implement specific delimiters or warnings to isolate transcribed text from instructions for the agent.
- Capability inventory: The script performs network operations to Alibaba Cloud APIs and third-party transcription URLs, and writes results to the local
output/directory. - Sanitization: Transcribed content and external metadata are processed and returned to the agent without filtering or escaping of potential embedded commands.
Audit Metadata