hf-cloud-sagemaker-deployment-planner

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill metadata and changelog document that the workflow was imported from Hugging Face's official GitHub repository (huggingface/skills).
  • [COMMAND_EXECUTION]: The skill instructs the agent to execute AWS CLI commands (aws service-quotas list-service-quotas) to verify account limits before recommending instance types, which is a legitimate and recommended operational practice.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input including model identifiers, S3 paths, and traffic descriptions to formulate deployment plans.
  • Ingestion points: User-provided model IDs, artifact locations, and workload descriptions entering via the discovery phase (SKILL.md).
  • Boundary markers: None explicitly defined for isolating user-provided strings within the internal planning logic.
  • Capability inventory: The skill coordinates subprocess calls for AWS CLI operations and references deployment scripts (deploy.py, deploy_async.py) used in downstream phases.
  • Sanitization: No explicit sanitization or validation logic is defined for the user-supplied model strings before they are incorporated into deployment recommendations.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 03:51 AM
Security Audit — agent-trust-hub — hf-cloud-sagemaker-deployment-planner