alibabacloud-dts-task-manager
Pass
Audited by Gen Agent Trust Hub on Apr 3, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill interacts with Alibaba Cloud services by executing
aliyunCLI commands. It includes comprehensive instructions for validating user input using regular expressions and sanitizing parameters to prevent shell injection by wrapping values in single quotes and rejecting illegal characters. It also mandates the masking of sensitive credentials in all user-facing output. - [EXTERNAL_DOWNLOADS]: The documentation guides users to install the
aliyunCLI from official vendor sources, specifically GitHub and Homebrew, which are trusted distribution channels. - [PROMPT_INJECTION]: The skill processes external data from user inputs and cloud API responses, creating a surface for indirect prompt injection.
- Ingestion points: User-provided values for database configuration (IPs, names, credentials) and JSON metadata returned from Alibaba Cloud APIs (e.g., DescribeDtsJobs).
- Boundary markers: The instructions require all user parameters to be wrapped in single quotes and sensitive values to be replaced with masked placeholders.
- Capability inventory: The agent has the capability to execute shell commands via the
aliyunCLI across all operational files (e.g., create-task.md, delete-task.md). - Sanitization: The skill defines robust validation logic, including regex checks for IP addresses and ports, and explicit rejection of shell metacharacters to ensure that external data is safely handled before being passed to the command line.
Audit Metadata