huawei-cloud-ces-elb-monitoring

Pass

Audited by Gen Agent Trust Hub on Sep 30, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPRIVILEGE_ESCALATIONPERSISTENCEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Downloads the official Huawei Cloud CLI (KooCLI) installation script and binaries from authorized vendor domains (e.g., myhuaweicloud.com).
  • [REMOTE_CODE_EXECUTION]: Provides instructions to execute the official vendor-authorized installation script to set up the CLI environment. These resources are verified as part of the official toolset for Huawei Cloud administration.
  • [COMMAND_EXECUTION]: Executes hcloud commands to query ELB metrics; the skill includes explicit security guidelines to ensure Access Keys (AK/SK) are never printed, requested, or stored within the conversation history.
  • [PRIVILEGE_ESCALATION]: Uses sudo within the installation guide to move the CLI binary to standard system execution paths (/usr/local/bin), which is standard behavior for administrative tool setup.
  • [PERSISTENCE]: Recommends adding auto-completion commands to user shell profiles (.bashrc, .zshrc) to improve CLI usability, which is a common and legitimate practice for development tools.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied resource IDs (SKILL.md) and ingests metadata from Cloud API responses (related-commands.md). While this provides a potential surface for indirect injection, the skill utilizes structured output templates (output-format.md) to manage data presentation. The capability inventory is restricted to administrative monitoring commands, and no explicit sanitization logic for API metadata is defined.
  • [SAFE]: Static detections for destructive commands were manually reviewed and identified as standard uninstallation and cache cleaning routines (e.g., removing the .hcloud directory) for the official CLI tool.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 30, 2026, 07:58 AM