
gcloud CLI Guardrails
Makes agents check gcloud command syntax, limit output, and avoid risky cloud actions
- Repository
- google/skills
- Stars
- 21,782
- Licence
- Apache-2.0
- Files
- 4 files
What it does
When an AI agent is asked to plan, write, or run Google Cloud command-line (gcloud) commands, this skill makes it follow a strict safety routine. The agent must look up the exact help text for each command before using it, instead of relying on memory, and must keep query results small by using filters, limits, and field selection. It also adds rules for denied operations, which require the user's explicit approval before the agent runs them.
Great for
- Developers who let an AI agent operate Google Cloud projects through the gcloud terminal tool
- Platform or ops engineers who want fewer hallucinated flags and fewer accidental destructive changes
- Teams that want consistent usage attribution tags on agent-run gcloud commands
Skip it if
- Writing Google Cloud client library code or raw REST/gRPC requests, which the skill's description explicitly excludes
- Situations where the agent has production credentials and no human review, because the skill relies on the agent following its rules rather than on hard technical enforcement
Try asking
- “Ask the agent to list my Cloud Run services in project my-project, showing only name and region”
- “Plan how to create a Cloud SQL instance and show the gcloud steps with validation first”
- “Explain the correct flags for gcloud compute instances create before running anything”
Install in 3 steps
- 1
Download the skill
The ZIP contains the 4 original files, the Apache-2.0 licence and an attribution note.
gcloud.zip - 2
Add it to your agent
Claude Code (all projects)
mkdir -p ~/.claude/skills && unzip gcloud.zip -d ~/.claude/skills/Claude Code / Agent SDK (this project only)
mkdir -p .claude/skills && unzip gcloud.zip -d .claude/skills/Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.
- 3
Ask for the task
No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.
Get better results
- Be specific about the project ID and the exact task so the agent can run the correct gcloud help lookups
- Review any command the agent proposes before approving it, especially those touching billing, IAM, or deletion
- Ensure the agent runs with least-privilege credentials, since the skill's guardrails are instructions rather than enforced controls
What's inside
- references/attribution.md
- references/cli-usage.md
- references/mcp-usage.md
- SKILL.md
Redistributed unmodified from google/skills under the Apache-2.0 licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.