storage-resize

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFECOMMAND_EXECUTIONPRIVILEGE_ESCALATIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides instructions to execute several system-level commands including growpart, xfs_growfs, resize2fs, btrfs, lsblk, df, and aliyun. These are necessary for managing disk partitions and file systems as described in the skill's purpose.
  • [PRIVILEGE_ESCALATION]: Modifications to the system partition table and file systems require root privileges. The skill documentation explicitly notes this requirement and includes yum install -y commands which require administrative access to the package manager.
  • [EXTERNAL_DOWNLOADS]: The skill instructs the agent to install standard system utilities (cloud-utils-growpart, xfsprogs, e2fsprogs, btrfs-progs) from the default system package repositories using yum. It also fetches instance metadata from the cloud provider's well-known internal metadata service at 100.100.100.200.
  • [INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it processes data from external system sources and cloud APIs.
  • Ingestion points: Data is ingested from the Alibaba Cloud metadata service (100.100.100.200) and the output of system tools like lsblk, df, and the aliyun CLI (e.g., in SKILL.md, examples.md, and reference.md).
  • Boundary markers: No explicit boundary markers or "ignore embedded instructions" warnings are present in the scripts when interpolating or processing this data.
  • Capability inventory: The skill utilizes high-privilege capabilities including package installation (yum), cloud resource modification (aliyun ecs ResizeDisk), and partition/file system writes (growpart, xfs_growfs, resize2fs).
  • Sanitization: External data is processed using standard shell utilities like awk, cut, grep, and tr to extract specific fields, but no dedicated validation for malicious instruction content is performed before the data is used to inform further actions.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 03:58 AM