huawei-cloud-apig-instance-management

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill provides instructions to download and execute a setup script for the Huawei Cloud CLI using a piped shell command (curl | bash). This operation targets the official Huawei Cloud download domain (myhuaweicloud.com).
  • [COMMAND_EXECUTION]: The skill relies on executing hcloud CLI commands to perform administrative tasks on cloud infrastructure, including the creation, modification, and deletion of dedicated API Gateway instances and associated policies.
  • [DATA_EXFILTRATION]: An included Python script (skill_quality_sdk.py) reports execution status, error messages, and masked metadata to a vendor-owned quality management API (myhuaweicloud.com). The script implements regex-based data masking to prevent the accidental transmission of sensitive information like API keys or passwords.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided parameters to construct CLI commands, presenting a surface for indirect prompt injection.
  • Ingestion points: User-provided inputs for instance IDs, names, and environment configurations in SKILL.md.
  • Boundary markers: The documented workflow requires explicit user confirmation before any mutating (R1/R2) commands are executed.
  • Capability inventory: The skill uses the hcloud CLI to perform full lifecycle management of cloud resources, including deletion.
  • Sanitization: The skill instructions do not explicitly define input sanitization routines, relying on the agent's platform-level guardrails.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 06:44 AM
Security Audit — agent-trust-hub — huawei-cloud-apig-instance-management