setting-up-workspaces-applications
Pass
Audited by Gen Agent Trust Hub on Nov 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- Infrastructure Provisioning via Command Execution: The skill orchestrates the creation of AWS AppStream 2.0 resources (fleets, stacks, image builders) by generating and executing AWS CLI commands. This is the primary function of the skill and is protected by a 'plan-first' requirement where the user must explicitly approve all actions before they are executed.
- External Software Downloads and Integrity Verification: For the sample workflow, the skill downloads applications such as GnuCash and LibreOffice from well-known official repositories (GitHub, Document Foundation). These downloads are performed within the isolated AppStream environment and are verified using hardcoded SHA-256 hashes to prevent the execution of tampered binaries.
- Sensitive Data Handling (Screenshots and Recordings): The skill configures the capture of desktop screenshots and session recordings for auditing. It proactively addresses the associated security risks by providing clear instructions on S3 bucket encryption, access control (IAM policies and Block Public Access), and server access logging.
- Least Privilege and Resource Isolation: Provisions are made to use IAM roles (via IamRoleArn) for infrastructure and agents, prioritizing fine-grained permissions over long-lived credentials. It also handles network discovery to avoid silent creation of insecure networking infrastructure.
- Transparency and User Oversight: The skill is designed with an 'adaptive' onboarding flow that translates technical jargon into plain language and requires explicit confirmation for cost-incurring or destructive operations.
Audit Metadata