2nd/10_Wiki/Topics/Architecture/G-Stack-Integration-Guide.md

id, title, category, status, canonical_id, aliases, duplicate_of, source_trust_level, confidence_score, verification_status, tags, raw_sources, last_reinforced, github_commit, tech_stack

id	title	category	status	canonical_id	aliases	duplicate_of	source_trust_level	confidence_score	verification_status	tags	raw_sources	last_reinforced	github_commit	tech_stack
wiki-2026-0508-g-stack-integration-guide	G Stack Integration Guide	10_Wiki/Topics	verified	self	G-Stack G Stack Integration GitHub-Gemini-Google Stack	none	A	0.9	applied	integration devops ci-cd ai google		2026-05-10	pending	language framework python github-actions

G-Stack Integration Guide

매 한 줄

"매 GitHub + Gemini + Google Cloud 를 single coherent dev stack 으로 묶는다". 매 G-Stack은 source control(GitHub), AI assist(Gemini Code Assist), cloud runtime(GCP/Cloud Run/Vertex AI) 의 통합 — 2026 Google ecosystem 의 매 default flow. GitHub Actions ↔ Cloud Build ↔ Vertex AI ↔ Gemini API.

매 핵심

매 G-Stack 구성

• GitHub: source, Actions CI/CD, Codespaces, Copilot 대안 = Gemini Code Assist
• Gemini: API (Gemini 2.5 Pro, Flash), Code Assist IDE plugin, Vertex AI
• Google Cloud: Cloud Run, GKE, Cloud Build, Artifact Registry, Vertex AI

매 핵심 integration points

• OIDC: 매 GitHub Actions → GCP keyless auth (no JSON key)
• Workload Identity Federation: 매 short-lived token
• Cloud Build trigger: 매 GitHub push → automated build
• Vertex AI agent: 매 Gemini 모델 + custom data RAG

매 응용

1. CI/CD: GitHub Actions deploy to Cloud Run.
2. AI-assisted dev: Gemini Code Assist in VSCode.
3. Custom RAG: Vertex AI Agent Builder + GitHub repo source.
4. Production LLM: Gemini API + Cloud Run wrapper.

💻 패턴

GitHub Actions → Cloud Run (OIDC, no key)

# .github/workflows/deploy.yml
name: Deploy to Cloud Run

on:
  push:
    branches: [main]

permissions:
  contents: read
  id-token: write  # 매 OIDC token 발급

jobs:
  deploy:
    runs-on: ubuntu-latest
    steps:
    - uses: actions/checkout@v4
    
    - id: auth
      uses: google-github-actions/auth@v2
      with:
        workload_identity_provider: projects/123456/locations/global/workloadIdentityPools/github/providers/github
        service_account: deploy@my-project.iam.gserviceaccount.com
    
    - uses: google-github-actions/setup-gcloud@v2
    
    - name: Build and Deploy
      run: |
        gcloud builds submit --tag us-central1-docker.pkg.dev/my-project/repo/app
        gcloud run deploy app \
          --image us-central1-docker.pkg.dev/my-project/repo/app \
          --region us-central1 \
          --allow-unauthenticated

Workload Identity Federation 설정 (Terraform)

resource "google_iam_workload_identity_pool" "github" {
  workload_identity_pool_id = "github"
}

resource "google_iam_workload_identity_pool_provider" "github" {
  workload_identity_pool_id          = google_iam_workload_identity_pool.github.workload_identity_pool_id
  workload_identity_pool_provider_id = "github"
  attribute_mapping = {
    "google.subject"       = "assertion.sub"
    "attribute.repository" = "assertion.repository"
  }
  attribute_condition = "assertion.repository_owner == 'myorg'"
  oidc {
    issuer_uri = "https://token.actions.githubusercontent.com"
  }
}

resource "google_service_account_iam_member" "github_act_as" {
  service_account_id = google_service_account.deploy.name
  role               = "roles/iam.workloadIdentityUser"
  member             = "principalSet://iam.googleapis.com/${google_iam_workload_identity_pool.github.name}/attribute.repository/myorg/myrepo"
}

Gemini API (Python, Cloud Run)

import os
from google import genai
from fastapi import FastAPI

client = genai.Client(api_key=os.environ["GEMINI_API_KEY"])
app = FastAPI()

@app.post("/chat")
async def chat(prompt: str):
    response = client.models.generate_content(
        model="gemini-2.5-pro",
        contents=prompt,
        config={
            "temperature": 0.7,
            "max_output_tokens": 2048,
        }
    )
    return {"text": response.text}

Vertex AI RAG (GitHub repo as source)

from google.cloud import aiplatform
from vertexai.preview import rag

aiplatform.init(project="my-project", location="us-central1")

corpus = rag.create_corpus(
    display_name="github-repo-rag",
    embedding_model_config=rag.EmbeddingModelConfig(
        publisher_model="publishers/google/models/text-embedding-005"
    )
)

# 매 GitHub mirror → GCS → Vertex
rag.import_files(
    corpus_name=corpus.name,
    paths=["gs://my-bucket/github-mirror/"],
    chunk_size=1024,
)

# 매 query
response = rag.retrieval_query(
    rag_resources=[rag.RagResource(rag_corpus=corpus.name)],
    text="How does the auth module work?",
    similarity_top_k=5,
)

Gemini Code Assist (VSCode settings)

{
  "geminicodeassist.project": "my-gcp-project",
  "geminicodeassist.enableInlineCompletions": true,
  "geminicodeassist.enableTelemetry": false,
  "github.copilot.enable": { "*": false }
}

Cloud Build trigger (GitHub push)

# cloudbuild.yaml
steps:
- name: 'gcr.io/cloud-builders/docker'
  args: ['build', '-t', 'us-central1-docker.pkg.dev/$PROJECT_ID/repo/app:$COMMIT_SHA', '.']
- name: 'gcr.io/cloud-builders/docker'
  args: ['push', 'us-central1-docker.pkg.dev/$PROJECT_ID/repo/app:$COMMIT_SHA']
- name: 'gcr.io/google.com/cloudsdktool/cloud-sdk'
  entrypoint: gcloud
  args:
  - run
  - deploy
  - app
  - --image=us-central1-docker.pkg.dev/$PROJECT_ID/repo/app:$COMMIT_SHA
  - --region=us-central1
options:
  logging: CLOUD_LOGGING_ONLY

Secret Manager → Cloud Run

# 매 secret 생성
echo -n "$GEMINI_KEY" | gcloud secrets create gemini-api-key --data-file=-

# 매 Cloud Run 에 마운트
gcloud run deploy app \
  --image=... \
  --update-secrets=GEMINI_API_KEY=gemini-api-key:latest \
  --service-account=deploy@my-project.iam.gserviceaccount.com

Monitoring (Cloud Logging + Sentry)

import google.cloud.logging
import sentry_sdk

google.cloud.logging.Client().setup_logging()
sentry_sdk.init(dsn=os.environ["SENTRY_DSN"], traces_sample_rate=0.1)

import logging
logging.info("매 structured log to Cloud Logging")

매 결정 기준

상황	G-Stack tool
Solo prototype	GitHub + Gemini Code Assist + Cloud Run
Production API	+ Vertex AI + Secret Manager + Cloud Build
ML/LLM heavy	Vertex AI Agent Builder + RAG
Enterprise	+ WIF + Org policy + VPC-SC
Multi-cloud	GitHub Actions abstraction layer

기본값: 매 OIDC (no JSON key), Cloud Run (serverless), Gemini 2.5 Flash (cheap default).

🔗 Graph

• 부모: DevOps · Cloud Architecture
• 변형: AWS Stack · Azure Stack · Vercel Stack
• 응용: GitHub Actions · Cloud Run · Vertex AI · Gemini API
• Adjacent: OIDC · Workload Identity · Infrastructure as Code

🤖 LLM 활용

언제: 매 GCP+GitHub 통합 troubleshooting, OIDC 설정 검증, Vertex AI agent 설계. 언제 X: 매 multi-cloud agnostic — G-Stack 은 GCP-tied.

❌ 안티패턴

• JSON service account key: 매 long-lived key — leak risk. OIDC 로 교체.
• Hardcoded Gemini key in repo: 매 obvious leak. Secret Manager 사용.
• Public Cloud Run: 매 --allow-unauthenticated 인데 매 sensitive endpoint → 매 IAM/IAP.
• No budget alert: 매 Vertex AI 무한 query → 매 unexpected bill.

🧪 검증 / 중복

• Verified (GitHub Docs, "Configuring OpenID Connect in Google Cloud Platform").
• Verified (Google Cloud Docs, Workload Identity Federation, 2024).
• Verified (Vertex AI RAG Engine GA, 2024).
• 신뢰도 A.

🕓 Changelog

날짜	변경
2026-05-08	Phase 1
2026-05-10	Manual cleanup — GitHub+Gemini+GCP integration patterns