alinux-cve-query

Warn

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: MEDIUMPRIVILEGE_ESCALATIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [PRIVILEGE_ESCALATION]: The skill instructs the agent to use sudo for administrative actions, specifically for updating system packages via sudo yum update --advisory <SA-ID> -y when a fix is available for a CVE.
  • [COMMAND_EXECUTION]: The skill executes several shell commands to manage the system environment, including cat /etc/os-release to identify the operating system, rpm -q to check installed package versions, and yum update to apply security advisories.
  • [EXTERNAL_DOWNLOADS]: The skill fetches CVE metadata and security advisory information from Alibaba Cloud's official security center API at https://alas.aliyuncs.com/api/portal/v1/cves/.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external data from a remote API and uses values from that data as arguments in system commands.
  • Ingestion points: Data is ingested from the Alibaba Cloud Security Center API (SKILL.md).
  • Boundary markers: None present; the agent interpolates external data directly into instructions.
  • Capability inventory: The skill has the capability to execute shell commands with administrative privileges (sudo yum, rpm, cat).
  • Sanitization: There is no evidence of sanitization or validation of the remote API response before it is used to construct shell commands.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Sep 14, 2026, 03:58 AM