cloud-iam-deep
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPRIVILEGE_ESCALATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill makes extensive use of official cloud CLI tools (aws, az, gcloud, kubectl) and curl to interact with cloud provider APIs and metadata services. These are standard tools for the skill's intended purpose of cloud auditing.
- [DATA_EXFILTRATION]: The instructions describe techniques for accessing sensitive cloud metadata endpoints (e.g., IMDS at 169.254.169.254) and fetching identity tokens or resource configurations. This behavior is documented as a mechanism for security auditors to identify exposed credentials and data.
- [PRIVILEGE_ESCALATION]: The skill documents over 40 specific privilege escalation vectors across major cloud providers (AWS, Azure, GCP). These are presented as a technical reference for identifying and testing for IAM misconfigurations during authorized engagements.
- [INDIRECT_PROMPT_INJECTION]: The skill identifies an attack surface where discovered cloud credentials (e.g., access keys found in code or metadata) are used as input for subsequent commands.
- Ingestion points: Discovered cloud credentials such as access keys, service account JSON files, and Kubernetes tokens (referenced in SKILL.md).
- Boundary markers: Not explicitly present for the ingestion of discovered credentials, though the skill provides extensive anti-patterns and usage guidelines.
- Capability inventory: Full use of cloud CLI tools and network operations via curl to official cloud provider domains (management.azure.com, graph.microsoft.com, googleapis.com).
- Sanitization: None present for the discovered credential data strings.
Audit Metadata