alibabacloud-qianwenai-support
Pass
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: SAFE
Full Analysis
- [COMMAND_EXECUTION]: The script
scripts/qianwen_support.pyexecutes theqianwenCLI tool to perform diagnostics and ticket operations. These executions are performed using argument lists without the shell, which prevents command injection risks. - [DATA_EXFILTRATION]: All network traffic is directed to official Alibaba Cloud / QianWen endpoints (
cli.qianwenai.com). The skill includes specific instructions and code logic to sanitize and mask sensitive information (e.g., API keys, phone numbers, and emails) before they are sent to the support platform. - [INDIRECT_PROMPT_INJECTION]: The skill acts as a bridge for user-generated issue descriptions. It mitigates the risk of processing malicious instructions within these descriptions by enforcing a mandatory 'preview and confirm' step, where users must validate the final ticket draft before it is submitted.
- [CREDENTIALS_UNSAFE]: Authentication is handled via the environment variable
QIANWEN_ACCESS_TOKENor the system's credential store (macOS keychain). The instructions explicitly forbid the output of plaintext credentials, ensuring that only the authentication status is visible during diagnostic checks.
Audit Metadata