aliyun-sms-bulk
Pass
Audited by Gen Agent Trust Hub on Mar 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides Python code templates using 'pandas' and 'openpyxl' to process local Excel files.
- [DATA_EXFILTRATION]: Instructions direct the agent to read and modify the IDE configuration file at '~/.cursor/mcp.json' and read local Excel files like '名单.xlsx'. While necessary for the skill's operation, accessing configuration files is a sensitive action.
- [EXTERNAL_DOWNLOADS]: The skill configures a remote MCP endpoint hosted on Alibaba Cloud ('aliyuncs.com'), which is a well-known service. This endpoint facilitates the SMS sending functionality.
- [PROMPT_INJECTION]: The skill processes untrusted data from Excel files, creating an indirect prompt injection surface. Evidence: 1. Ingestion points: '名单.xlsx' (SKILL.md); 2. Boundary markers: A mandatory '预览确认' (preview and confirm) step is defined in the workflow; 3. Capability inventory: SMS sending via SendSms, SendBatchSms, and SendMessageWithTemplate tools; 4. Sanitization: A human-in-the-loop confirmation step is required before the agent executes the bulk sending process.
Audit Metadata