alicloud-compute-ecs

Pass

Audited by Gen Agent Trust Hub on Mar 11, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill enables execution of shell or PowerShell commands on remote ECS instances through the official Alibaba Cloud Assistant (RunCommand) service. Evidence: scripts/run_remote_command.py uses the RunCommand API to send and poll command results.
  • [EXTERNAL_DOWNLOADS]: Fetches well-known Alibaba Cloud SDK packages from the official Python Package Index (PyPI) to support cloud operations. Evidence: SKILL.md setup instructions include pip install for alibabacloud_ecs20140526, alibabacloud_tea_openapi, alibabacloud_credentials, and alibabacloud_cms20190101.
  • [PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection by processing untrusted data from external cloud resources.
  • Ingestion points: scripts/run_remote_command.py (captures stdout from remote instances), scripts/list_instances_all_regions.py (processes instance names and metadata).
  • Boundary markers: None identified in instructions or scripts to delimit external content.
  • Capability inventory: Extensive cloud management capabilities including instance creation, deletion, and remote execution via the ECS SDK.
  • Sanitization: No explicit validation or sanitization is performed on data retrieved from the Alibaba Cloud API before inclusion in the agent context.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 11, 2026, 10:15 AM