メニュー
Google Cloud Run deployment, service management, traffic splitting, and log inspection. Use when deploying containerized apps to Cloud Run, managing services, or inspecting Cloud Run logs.
デフォルトでは、最初にソースを確認する Prompt が選択されています。直接コマンドに切り替えるか、ローカルコピーをダウンロードすることもできます。
インストールを決める前に、SKILL.md と SkillsMP に表示されている付属ファイルをお読みください。
Codex または Claude でインストール この Prompt をコピーして Codex、Claude、または他のアシスタントに貼り付けると、Skill ページを確認してインストールできます。
直接コマンドでは確認用 Prompt が省略されます。実行前にソースを確認してください。
npx skills add https://github.com/oimiragieo/agent-studio --skill cloud-runコマンドは1行のまま表示されます。コピー前に横へスクロールして全体を確認してください。
ローカルで確認しますか？SkillsMP が現在取得できるファイルをダウンロードできます。
Agent frontmatter enhancements — disallowedTools, mcpServers scoping, fork_eligible field
Context management — pre-compact persistence, threshold alignment, microcompact/circuit breaker detection
Hook enhancements — updatedInput for bash safety prefixes, suppressOutput for verbose blocks, denial-based routing feedback
SKILL.md を表示中
SOC 職業分類に基づく
| name | cloud-run |
| description | Google Cloud Run deployment, service management, traffic splitting, and log inspection. Use when deploying containerized apps to Cloud Run, managing services, or inspecting Cloud Run logs. |
| version | 1.0.0 |
| model | sonnet |
| invoked_by | both |
| user_invocable | true |
| tools | ["Read","Write","Bash","Grep"] |
| agents | ["devops","cloud-devops-expert"] |
| category | deployment |
| tags | ["gcloud","cloud-run","gcp","containers","serverless","deployment"] |
| best_practices | ["Always specify --region for all gcloud run commands","Use --allow-unauthenticated only for public services","Set memory and CPU limits explicitly","Use Cloud Build for production container builds"] |
| error_handling | graceful |
| streaming | supported |
| verified | false |
| source | builtin |
| trust_score | 100 |
| provenance_sha | b13db5d3a2eee6b1 |
Deploy and manage containerized applications on Google Cloud Run. Covers service deployment, revision management, traffic splitting, and log inspection.
gcloud) installed and authenticated# Deploy from source (Cloud Build auto-builds)
gcloud run deploy SERVICE_NAME \
--source . \
--region us-central1 \
--allow-unauthenticated
# Deploy from container image
gcloud run deploy SERVICE_NAME \
--image gcr.io/PROJECT_ID/IMAGE:TAG \
--region us-central1 \
--memory 512Mi \
--cpu 1 \
--min-instances 0 \
--max-instances 10
gcloud run services list --region us-central1 --format json
gcloud run services describe SERVICE_NAME \
--region us-central1 \
--format json
gcloud run services logs read SERVICE_NAME \
--region us-central1 \
--limit 100
# Tail logs in real-time
gcloud run services logs tail SERVICE_NAME \
--region us-central1
# Route 50% to latest, 50% to previous revision
gcloud run services update-traffic SERVICE_NAME \
--region us-central1 \
--to-revisions LATEST=50,REVISION_NAME=50
# Route all traffic to latest
gcloud run services update-traffic SERVICE_NAME \
--region us-central1 \
--to-latest
# Set env vars
gcloud run services update SERVICE_NAME \
--region us-central1 \
--set-env-vars KEY1=VALUE1,KEY2=VALUE2
# Mount secret as env var
gcloud run services update SERVICE_NAME \
--region us-central1 \
--set-secrets ENV_NAME=SECRET_NAME:latest
gcloud run services delete SERVICE_NAME \
--region us-central1 \
--quiet
Either use a Dockerfile or let Cloud Build auto-detect:
# Option A: Build and push manually
docker build -t gcr.io/PROJECT_ID/SERVICE_NAME:latest .
docker push gcr.io/PROJECT_ID/SERVICE_NAME:latest
# Option B: Deploy from source (Cloud Build handles it)
gcloud run deploy SERVICE_NAME --source .
gcloud run deploy SERVICE_NAME \
--image gcr.io/PROJECT_ID/SERVICE_NAME:latest \
--region us-central1 \
--port 8080 \
--memory 512Mi \
--cpu 1
# Get service URL
gcloud run services describe SERVICE_NAME \
--region us-central1 \
--format "value(status.url)"
# Test the endpoint
curl $(gcloud run services describe SERVICE_NAME \
--region us-central1 \
--format "value(status.url)")
gcloud run services logs read SERVICE_NAME --region us-central1 --limit 20
gcloud run commands — omitting region causes interactive prompts that break automation.--set-secrets flag.| Anti-Pattern | Why It Fails | Correct Approach |
|---|---|---|
| Missing --region flag | Interactive prompt breaks CI/CD | Always specify region explicitly |
| Public endpoints for internal APIs | Security exposure | Use IAM auth + Cloud Run invoker role |
| Hardcoded secrets in env vars | Secrets visible in service config | Use Secret Manager integration |
| No min-instances for latency-critical | Cold start adds 2-10s latency | Set --min-instances 1+ for critical paths |
| Single revision (no traffic split) | No rollback path | Always keep previous revision, split traffic gradually |
When the cloud-run-mcp MCP server is configured, these tools are available:
| Tool | Description | Key Params |
|---|---|---|
deploy-file-contents | Deploy code directly from file content | service_name, region, source_files |
list-services | List all Cloud Run services | project_id, region |
get-service | Get service details + URL | service_name, project_id, region |
get-service-log | Fetch recent logs | service_name, project_id, limit |
deploy-local-folder | Deploy from local directory | folder_path, service_name, region |
list-projects | List GCP projects | — |
create-project | Create new GCP project | project_id, project_name |
MCP setup: npx @google-cloud/cloud-run-mcp or configure in Claude Desktop settings.json
Auth: Application Default Credentials (gcloud auth application-default login)
When to use MCP vs CLI: Use MCP tools when user asks "deploy this code" or "what services are running"; use gcloud CLI for advanced operations (IAM, VPC, secrets).
Before starting:
Read .claude/context/memory/learnings.md
After completing:
.claude/context/memory/learnings.md.claude/context/memory/issues.md.claude/context/memory/decisions.mdASSUME INTERRUPTION: If it's not in memory, it didn't happen.