aws-best-practice-research

Pass

Audited by Gen Agent Trust Hub on Apr 16, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses the AWS CLI (aws) and jq to collect resource configuration data for assessment. These operations are restricted to read-only actions (e.g., describe-*, list-*) and are only triggered if the user explicitly provides credentials and resource identifiers.
  • [EXTERNAL_DOWNLOADS]: The skill documentation references and utilizes the aws-knowledge-mcp-server from the official awslabs GitHub repository to search and read AWS documentation. This is a well-known, trusted source for AWS-related tooling.
  • [DATA_EXFILTRATION]: While the skill reads sensitive resource configurations (topology, encryption status, auth methods) using the AWS CLI, this data is processed locally to generate a markdown report. There are no patterns indicating the data is sent to unauthorized external domains.
  • [PROMPT_INJECTION]: The skill instructions include clear boundary markers and specific logic for handling service scope (e.g., infra-level vs. workload-level for EKS). There are no instructions to bypass safety filters or override system constraints.
  • [PRIVILEGE_ESCALATION]: The skill instructions specifically state that only read-only IAM permissions are required for the target service and that write permissions are never needed.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 16, 2026, 06:06 AM