alibabacloud-polardb-ai-assistant
Fail
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: HIGHEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONREMOTE_CODE_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill includes instructions to download the Aliyun CLI installation script from a vendor-controlled CDN (aliyuncli.alicdn.com). This is a standard resource for this vendor.
- [REMOTE_CODE_EXECUTION]: The installation guide in SKILL.md documents a process to download and execute a shell script from a remote vendor source. The instructions include a manual safety checkpoint (reviewing the script with 'less') before execution.
- [COMMAND_EXECUTION]: The skill invokes the aliyun CLI and system utilities such as jq to perform database O&M tasks and parse responses.
- [INDIRECT_PROMPT_INJECTION]: The skill features a data ingestion surface that could be susceptible to indirect prompt injection from untrusted data processed at runtime.
- Ingestion points: Natural language queries and streaming backend API responses are processed in the scripts/call_yaochi_agent.sh script.
- Boundary markers: There are no explicit delimiters or boundary instructions in the prompt logic to isolate untrusted data.
- Capability inventory: The skill leverages the aliyun CLI to interact with cloud database APIs, providing significant management capabilities.
- Sanitization: The bash script includes validation for query length and session ID character format to prevent basic injection patterns.
Recommendations
- HIGH: Downloads and executes remote code from: https://aliyuncli.alicdn.com/setup.sh - DO NOT USE without thorough review
Audit Metadata