Deployment Manager

Overview

Deploying to production should be boring. Deployment Manager makes it boring by adding intelligence to your release process — predicting risk before you deploy, monitoring health during rollout, and automatically rolling back if something goes wrong.

The skill analyzes your code changes against historical deployment data to generate a risk score. Database migrations, critical path changes, and dependency updates score higher. Based on the risk level, it recommends an appropriate deployment strategy (direct, blue-green, canary) and manages the rollout.

During deployment, it monitors your configured health metrics (error rate, latency, CPU, memory). If metrics degrade beyond thresholds, it automatically rolls back — no human intervention needed.

How It Works

1. Analyze — Assess deployment risk based on code changes and history
2. Plan — Recommend deployment strategy based on risk level
3. Execute — Manage the deployment rollout (canary, blue-green, direct)
4. Monitor — Watch health metrics during and after deployment
5. Decide — Auto-promote or auto-rollback based on health checks

Use Cases

• Risk assessment — Know deployment risk before clicking deploy
• Canary management — Automated canary rollouts with metric-based promotion
• Rollback automation — Instant rollback when health degrades
• Deployment tracking — Complete audit trail of all deployments
• Release confidence — Build team confidence in the deployment process

Getting Started

# deploy-config.yaml
monitoring:
  error_rate_threshold: 1%
  latency_p95_threshold: 500ms
  check_interval: 30s
strategy:
  low_risk: direct
  medium_risk: blue-green
  high_risk: canary
canary:
  initial_percentage: 5
  increment: 15
  promotion_interval: 5m

Example

Deployment: api-service v2.14.0 → v2.15.0

Risk Assessment:
  📊 Risk Score: 62/100 (Medium)
  - Database migration detected (+20)
  - Authentication module changes (+15)
  - 847 lines changed across 12 files (+12)
  - All tests passing (-10)
  - Similar past deployment succeeded (-5)

Recommended Strategy: Canary Deployment

Rollout:
  14:00 — 5% canary deployed ✅
  14:05 — Health check: error rate 0.2%, latency P95 180ms ✅
  14:05 — Promoting to 20% ✅
  14:10 — Health check: error rate 0.3%, latency P95 195ms ✅
  14:10 — Promoting to 50% ✅
  14:15 — Health check: all clear ✅
  14:15 — Promoting to 100% ✅

Deployment complete. Confidence score: 94/100

Alternatives

• Argo Rollouts — Kubernetes progressive delivery controller
• Spinnaker — Multi-cloud continuous delivery platform
• LaunchDarkly — Feature flag management with gradual rollouts