The Agent Skills Directory

Command Execution: The skill generates and facilitates the execution of bash scripts and Google Cloud CLI commands for deployment. This is a core feature for infrastructure automation but requires user oversight to ensure the generated commands align with expectations.
Dynamic Execution and Script Generation: The agent dynamically creates Terraform HCL files and shell scripts based on user design specifications. This involves using chmod +x to make these generated scripts executable. These steps are standard for DevOps workflows but represent a security consideration as the agent is generating and running code at runtime.
External Module and Resource References: The skill utilizes official Google-managed Terraform modules from the Terraform Registry. These sources are considered trusted and are used to ensure best-practice configurations, though they involve fetching resources from external registries during the deployment process.
Infrastructure Resource Discovery: The skill uses gcloud commands to retrieve information about existing Google Cloud resources (like buckets and GKE clusters) to assist in configuration. This involves reading project metadata to provide accurate design options.
Indirect Prompt Injection Surface: The skill processes external resource names and project metadata which are interpolated into generated scripts. * Ingestion points: Google Cloud resource lists retrieved in references/resource-discovery.md. * Boundary markers: Not explicitly detailed in the logic. * Capability inventory: Includes shell command execution and infrastructure deployment actuation. * Sanitization: The skill enforces naming constraints for resource IDs in references/terraform-generation.md to ensure API compatibility.

google-cloud-global-frontend-configuration