The Agent Skills Directory

Command Execution (gcloud): The skill identifies target resources by executing gcloud commands such as gcloud run services list and gcloud asset search-all-resources. These commands are used within the agent's environment to discover existing cloud infrastructure, which helps automate the definition of the SLO service scope.
Indirect Prompt Injection Surface: As the skill ingests and processes data from external sources—specifically the output of system commands (gcloud) and public Google Cloud documentation—there is a potential surface for indirect prompt injection. While the sources are trusted, the agent processes these inputs to determine metrics and scopes.
Ingestion points: gcloud command results and external documentation URLs in SKILL.md and references/service_metrics.md.
Boundary markers: The instructions do not define explicit boundary markers for external data.
Capability inventory: The skill utilizes file writing (write_to_file) to output Terraform and system command execution (gcloud) for resource discovery.
Sanitization: Explicit sanitization or escaping of external content before interpolation is not detailed in the instructions.
External Resource Access: The skill references official Google Cloud documentation and GitHub issue tracking. These references are used to fetch up-to-date service metrics and provide user support, which is consistent with the skill's purpose as a cloud management tool.

google-cloud-slo-alert-configuration