Google Cloud CLI

gcloud - manage Google Cloud resources and developer workflows

This skill is built on top of the official gcloud CLI. It supports the full CLI surface while avoiding hardcoded syntax by always consulting --help output at runtime.

Related docs:

• Installation and setup: installation.md
• Group reference: groups.md
• Usage examples: examples.md
• Troubleshooting: troubleshooting.md

Requirements

This skill requires gcloud CLI.

For setup instructions, see installation.md.

Scope

Use this skill only for Google Cloud resource management via gcloud commands. Do not use unrelated endpoints, tools, or local file operations outside the requested task.

Credentials and Environment

This skill uses the active Google Cloud CLI authentication context (gcloud auth) and configuration (gcloud config). It inherits the permissions of the active identity.

Before any operation:

1. Run gcloud config list --format='text(core.account,core.project)' and show the active account and project.
2. If the active account is not a dedicated service account, stop and ask the user to switch identities.
3. Confirm the target project and environment with the user before proceeding.

Credential safety rules:

• Use least-privilege service accounts.
• Do not use personal accounts or broad admin identities for automation.
• Be explicit when --impersonate-service-account is in use.
• Prefer sandbox projects for validation before production changes.

Workflow

Before executing any gcloud command, follow this sequence:

1. Check active context:

   gcloud config list --format='text(core.account,core.project)'
   

2. Identify the right command group from groups.md.
3. Discover syntax using help commands:

   gcloud <GROUP> --help
   gcloud <GROUP> <SUBGROUP> --help
   

4. Build the exact command from discovered syntax.
5. Present the full command and wait for explicit user approval.
6. Execute only after approval.
7. Return output and summarize result.

Approval Policy

All operations require explicit user confirmation before execution, including read operations.

This includes:

• Read/list/get operations
• Create/update/delete operations
• IAM and policy changes
• Configuration changes (set, unset, reset)
• Service enable/disable operations

For every operation, the agent must:

1. Show the full command.
2. Show active account/project context.
3. Wait for explicit user approval.

Important Rules

• Never guess command syntax; always validate with --help first.
• Never execute commands autonomously.
• Use --format=json when output will be parsed programmatically.
• Use --quiet only after explicit user approval.
• Warn clearly when commands are high-impact (IAM, networking, deletion, org-level changes).

What You Can Do