aliyun-qwen-tts

Warn

Audited by Gen Agent Trust Hub on Apr 28, 2026

Risk Level: MEDIUMDATA_EXFILTRATIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [DATA_EXFILTRATION]: The script scripts/generate_tts.py is configured to read sensitive API credentials from the local file system at ~/.alibabacloud/credentials if not found in environment variables.
  • [DATA_EXFILTRATION]: In scripts/generate_tts.py, the call_generate function permits the API's base URL to be overridden by the input request (req.get("base_url", ...)). If the skill processes untrusted data that specifies a malicious URL, the DASHSCOPE_API_KEY (harvested from the environment or credentials file) would be transmitted to that unauthorized external server.
  • [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it ingests potentially untrusted JSON data via the --request or --file arguments in scripts/generate_tts.py without boundary markers or sanitization, and uses this data to influence network requests.
  • [EXTERNAL_DOWNLOADS]: The skill fetches the dashscope package from the official Python package registry, which is the legitimate SDK for Alibaba Cloud's Model Studio.
  • [COMMAND_EXECUTION]: The documentation and scripts involve executing shell commands for package management (pip install), script compilation (py_compile), and running the TTS generation utility.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 28, 2026, 03:35 AM