Cloud Run Deployment Skill

Target Token Efficiency: 65% (350 tokens -> 120 tokens)

Purpose

Automate Cloud Run AI Engine deployment with free tier optimized configuration, without manual script explanation.

Trigger Keywords

•"deploy cloud run"
•"cloud run deploy"
•"deploy ai engine"
•"ai engine deploy"
•"production deploy"
•"배포"
•"클라우드 런 배포"
•"AI 엔진 배포"
•"프로덕션 배포"
•"gcp deploy"

Context

•Project: OpenManager AI v8.0.0
•Region: asia-northeast1 (Seoul)
•Service: ai-engine
•Tier: FREE TIER OPTIMIZED
•Free Limits: 180,000 vCPU-sec, 360,000 GB-sec, 2M requests/month

Configuration (Free Tier)

Cloud Run

Resource	Value	Free Hours/Month
CPU	1 vCPU	~50 hours
Memory	512Mi	~200 hours
Max Instances	3	-
Concurrency	80	-
Timeout	300s	-

Cloud Build

Resource	Value	Free Limit
Machine	default (e2-medium)	120 min/day

⚠️ FREE TIER RULES (절대 위반 금지):

•deploy.sh에 --machine-type 옵션 추가 금지

•cloudbuild.yaml에 machineType 설정 추가 금지

•E2_HIGHCPU_8, N1_HIGHCPU_8 등 커스텀 머신은 무료 대상 아님

•기본 머신(e2-medium)만 일 120분 무료 빌드 제공

Workflow

1. Pre-deployment Check

bash

# Check current working directory
cd /mnt/d/cursor/openmanager-ai/cloud-run/ai-engine

# Verify GCP project
gcloud config get-value project

# Check for uncommitted changes
git status --short

Expected:

•Project: openmanager-free-tier
•Clean working tree (recommended)

2. TypeScript Build Check

bash

# Verify TypeScript compiles
npm run build --prefix /mnt/d/cursor/openmanager-ai/cloud-run/ai-engine

Expected:

•No compilation errors
•dist/ directory updated

3. Execute Deployment

bash

# Run deployment script
cd /mnt/d/cursor/openmanager-ai/cloud-run/ai-engine && bash deploy.sh

Deployment Steps (automated by script):

•Build container image (Cloud Build, ~2-3 min)
•Deploy to Cloud Run (~1 min)
•Health check
•Cleanup old images/revisions (background)

4. Verify Deployment

bash

# Health check
curl -s https://ai-engine-490817238363.asia-northeast1.run.app/health | jq

# Check monitoring endpoint
curl -s https://ai-engine-490817238363.asia-northeast1.run.app/monitoring | jq '.agents'

Expected Health Response:

json

{
  "status": "healthy",
  "version": "8.0.x",
  "agents": {
    "active": 4,
    "available": ["NLQ", "Analyst", "Reporter", "Advisor"]
  }
}

Report Summary Format

code

🚀 Cloud Run Deployment Results
├─ Service: ai-engine
├─ Region: asia-northeast1
├─ Version: {SHORT_SHA}
├─ URL: https://ai-engine-*.run.app
├─ Status: ✅ Success / ❌ Failed
│
├─ Resource Config (FREE TIER):
│   ├─ CPU: 1 vCPU (~50 hrs/month free)
│   ├─ Memory: 512Mi (~200 hrs/month free)
│   └─ Max Instances: 3
│
└─ Health Check: ✅ Pass / ❌ Fail

Token Optimization Strategy

Before (Manual):

code

User: "배포해줘"
Assistant: [reads Dockerfile, explains config, explains free tier, runs commands, checks logs]
Tokens: ~350

After (Skill):

code

User: "deploy cloud run"
Skill: [executes workflow, reports summary]
Tokens: ~120 (65% reduction)

Edge Cases

배포 문제 해결: references/edge-cases.md

•Build Failure
•Deployment Timeout
•Health Check Failure
•Permission Denied
•Free Tier 초과

Rollback

롤백 절차: references/rollback.md

빠른 롤백:

bash

# 이전 리비전으로 트래픽 전환
gcloud run services update-traffic ai-engine \
  --region asia-northeast1 \
  --to-revisions REVISION_NAME=100

Related Files

File	Purpose
`cloud-run/ai-engine/Dockerfile`	Multi-stage build config
`cloud-run/ai-engine/deploy.sh`	Deployment script
`cloud-run/ai-engine/cloudbuild.yaml`	CI/CD config (optional)

Success Criteria

•Build: Completed without errors
•Deploy: Service updated successfully
•Health: HTTP 200 from /health
•Agents: All 5 agents active

Related Skills

•lint-smoke - Pre-deployment code quality check

Changelog

•
2026-02-02: v1.1.0 - Cloud Build free tier optimization
- •Removed machineType: E2_HIGHCPU_8 from cloudbuild.yaml
- •Removed --machine-type=e2-highcpu-8 from deploy.sh
- •Added free tier guard rules to prevent recurrence
•
2026-01-06: v1.0.0 - Initial implementation
- •Free tier optimized configuration
- •1 vCPU, 512Mi memory, 3 max instances
- •Integrated health check verification
- •10 trigger keywords (Korean/English)