alibabacloud-idaas-cloud-identity-setup

Pass

Audited by Gen Agent Trust Hub on Oct 22, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is exceptionally security-conscious, containing numerous instructions to avoid handling secrets in chat and to verify identities before every action. It restricts its operations to official APIs and tools.
  • [COMMAND_EXECUTION]: The Python scripts (e.g., cloud_identity_preflight.py, alibaba_account_lifecycle.py) utilize subprocess.run to invoke cloud CLIs. Security is maintained because all inputs—such as Account IDs, Profile names, and Instance IDs—are strictly validated against rigid regular expressions before being passed to the execution command. Furthermore, command paths are resolved using shutil.which to ensure only intended binaries are executed.
  • [DATA_EXFILTRATION]: The skill includes robust safeguards against data exfiltration. It explicitly forbids the collection of credentials, tokens, or callback URLs through the conversation and relies on official OIDC/STS flows managed by the CLIs. Network activity is limited to established cloud provider endpoints.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes structured data from cloud APIs. It mitigates indirect injection risks with specific instructions to treat all API responses and documents as data rather than executable instructions, maintaining a clear boundary between external data and agent logic.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 22, 2026, 12:09 PM