azurecli
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill utilizes the Azure CLI (
az) to perform cloud management tasks. It includes examples for creating AKS clusters, managing resource groups, and deploying Bicep templates. These commands are typical for the skill's intended use case. - [INDIRECT_PROMPT_INJECTION]: The skill processes data from external cloud environments through command outputs (e.g.,
az vm list,az webapp show). This presents a potential attack surface where instructions embedded in resource metadata (like VM names or tags) could be ingested by the agent. - Ingestion points: Output from
az vm list,az network public-ip show, andaz webapp showmentioned inSKILL.md. - Boundary markers: None present; the instructions do not specify delimiters for tool outputs.
- Capability inventory: Shell command execution for Azure resource management and credential retrieval.
- Sanitization: No explicit sanitization or filtering of command output is defined in the instructions.
Audit Metadata