Cloud Penetration Testing
Fail
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: HIGHREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONPRIVILEGE_ESCALATIONPERSISTENCEINDIRECT_PROMPT_INJECTION
Full Analysis
- [REMOTE_CODE_EXECUTION]: The installation procedure for the Google Cloud SDK utilizes a direct pipe from curl to bash (
curl https://sdk.cloud.google.com | bash). While this is a standard installation method provided by a well-known service, it represents a remote code execution pattern. - [EXTERNAL_DOWNLOADS]: The skill downloads several CLI tools and frameworks from external sources, including the AWS CLI from Amazon's official servers and various security tools like ScoutSuite and Pacu via pip.
- [DATA_EXFILTRATION]: Instructions include commands for extracting sensitive data from cloud environments, such as dumping Azure Key Vault secrets (
az keyvault secret show), extracting Lambda environment variables, and exporting Azure authentication contexts to local files (Save-AzContext). - [PRIVILEGE_ESCALATION]: The skill provides techniques for escalating privileges within cloud tenants, including assigning service principals to the 'Global Administrator' (Company Administrator) role and creating new administrative users.
- [PERSISTENCE]: Detailed methods for establishing long-term access are documented, such as creating backdoor service principals in Azure AD and generating secondary IAM access keys in AWS.
- [INDIRECT_PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection as it processes untrusted external data.
- Ingestion points: Reads from user-provided files such as
users.txt,ips.txt, andpasslist.txtinSKILL.mdandreferences/advanced-cloud-scripts.md. - Boundary markers: None identified; data is processed directly via shell loops and PowerShell scripts.
- Capability inventory: Significant capabilities including
curl,Invoke-AzVMRunCommand(executing code on remote VMs), and broad file system write access via output redirection. - Sanitization: No evidence of input validation or escaping for the contents of the ingested files before they are used in command arguments.
- [COMMAND_EXECUTION]: The skill heavily relies on the execution of shell and PowerShell commands to perform its functions, including interaction with VM management APIs (
Invoke-AzVMRunCommand).
Recommendations
- HIGH: Downloads and executes remote code from: https://sdk.cloud.google.com - DO NOT USE without thorough review
Audit Metadata