🤖 Deploy Code lên Zeni Cloud — 1 API Call

Hướng dẫn cho AI Agents (Claude Sessions, ChatGPT, Cursor, Replit Agents). Một endpoint POST /deploy/quick để deploy bất kỳ app nào lên production trong 60-90 giây.

🚀 Bắt đầu — 30 giây

Bước 1: Lấy ZENI_TOKEN

Login dashboard zenicloud.io/app → Tab API Tokens → "+ New Token" → copy token bắt đầu bằng zeni_pat_

Bước 2: Chọn 1 trong 3 cách deploy

curl -X POST "https://zenicloud.io/api/v1/deploy/quick?ws=$WORKSPACE" \
  -H "Authorization: Bearer $ZENI_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "image": "docker.io/library/nginx:alpine",
    "name": "my-static-site",
    "port": 80
  }'

{
  "deploy_id": "...",
  "method": "docker_image",
  "status": "deploying",
  "project_id": "uuid",
  "poll_url": "https://zenicloud.io/api/v1/projects/{id}?ws=...",
  "estimated_live_in_sec": 30
}

# Bash + Python (POSIX)
ZIP_B64=$(zip -qr - ./my-app/ | base64 -w0)
curl -X POST "https://zenicloud.io/api/v1/deploy/quick?ws=$WORKSPACE" \
  -H "Authorization: Bearer $ZENI_TOKEN" \
  -H "Content-Type: application/json" \
  -d "{\"zip_base64\":\"$ZIP_B64\",\"name\":\"my-app\",\"framework\":\"auto\"}"

import requests, zipfile, io, base64

buf = io.BytesIO()
with zipfile.ZipFile(buf, 'w', zipfile.ZIP_DEFLATED) as z:
    z.writestr('index.html', '<h1>Hello World</h1>')
    z.writestr('package.json', '{"name":"my-app"}')

r = requests.post(
    'https://zenicloud.io/api/v1/deploy/quick',
    params={'ws': 'your-workspace'},
    headers={'Authorization': 'Bearer zeni_pat_xxx'},
    json={
        'zip_base64': base64.b64encode(buf.getvalue()).decode(),
        'name': 'my-app',
        'framework': 'auto',  # auto-detect from package.json/requirements.txt
        'port': 8080
    }
)
print(r.json())
# → {"deploy_id":"...", "poll_url":"...", "status":"queued"}

curl -X POST "https://zenicloud.io/api/v1/deploy/quick?ws=$WORKSPACE" \
  -H "Authorization: Bearer $ZENI_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "repo_url": "https://github.com/owner/repo",
    "branch": "main",
    "name": "my-app",
    "framework": "auto"
  }'

🎯 Auto-detect Framework

Zeni Cloud tự nhận diện framework từ files trong ZIP/repo:

Override bằng "framework": "nextjs|react|vue|fastapi|express|static"

📊 Poll Deploy Status

curl "$POLL_URL" -H "Authorization: Bearer $ZENI_TOKEN"

# Response:
{
  "deploy_id": "...",
  "status": "building",       # queued|extracting|building|deploying|success|failed
  "name": "my-app",
  "framework": "express",
  "image_url": null,          # populated khi build SUCCESS
  "deploy_url": null,         # populated khi deploy SUCCESS
  "error": null,
  "completed_at": null
}

# Khi success:
{
  "deploy_id": "...",
  "status": "success",
  "image_url": "us-central1-docker.pkg.dev/zeni-cloud-core/zeni-images/zeni-myws-myapp:abc123",
  "deploy_url": "https://my-app-xxx-uc.a.run.app",
  "completed_at": "2026-05-03T..."
}

🛠 Working Examples for AI Agents

Claude Code Session (clone repo + deploy)

// Pseudocode for Claude Code:
1. User: "Deploy my Next.js app at /home/user/my-app to Zeni Cloud"
2. Bash: cd /home/user/my-app && zip -qr /tmp/app.zip ./
3. Python: read /tmp/app.zip, base64 encode
4. curl POST /deploy/quick with zip_base64
5. Loop: poll /deploy/quick/{id} every 10s until status='success'
6. Return deploy_url to user

ChatGPT Code Interpreter

import requests, zipfile, base64
from pathlib import Path

# Upload + deploy in one cell
def deploy_to_zeni(folder_path, ws, token, name=None, framework='auto'):
    buf = io.BytesIO()
    with zipfile.ZipFile(buf, 'w', zipfile.ZIP_DEFLATED) as z:
        for f in Path(folder_path).rglob('*'):
            if f.is_file() and 'node_modules' not in str(f) and '.git' not in str(f):
                z.write(f, f.relative_to(folder_path))

    r = requests.post(f'https://zenicloud.io/api/v1/deploy/quick',
        params={'ws': ws},
        headers={'Authorization': f'Bearer {token}'},
        json={'zip_base64': base64.b64encode(buf.getvalue()).decode(),
              'name': name or folder_path.split('/')[-1], 'framework': framework})
    return r.json()

Cursor Composer / Replit Agent

Same pattern: zip current workspace → POST /deploy/quick → poll → return URL.

💡 Tips for AI Agents

• Always use framework: "auto" trừ khi user explicit override
• Default region: asia-southeast1 (Singapore — gần VN nhất)
• Skip node_modules/, .git/, __pycache__/ khi zip — giảm size
• Poll status max 90s — nếu vẫn building sau 90s, tell user check dashboard
• Catch 413 Payload Too Large — ZIP > 50MB → tell user remove large files
• Catch 401 → token invalid, tell user regenerate at /app#api-tokens
• Catch 404 trên poll → deploy_id mismatch hoặc khác workspace

📋 Endpoint Reference

🤖 Designed for AI agents. Tối giản, idempotent, single-call. Built by Zeni Cloud.