Backup & Restore

Overview

MoltGhost's automated backup system preserves complete agent state—configuration, memory, skills, and runtime data.

Point-in-time recovery ensures agents resume exactly where they left off, even after termination or failure.

Agent State → Incremental Backup → S3 Storage → Restore → Identical Agent

What Gets Backed Up

Complete Agent Snapshot (daily automated + manual):

Total Backup Size: 1MB (chat agent) → 500MB+ (enterprise w/ long memory)

Backup Architecture

┌─────────────────────────────────────────────────────────────┐
│                       Agent Pod                             │
├─────────────────────────────────────────────────────────────┤
│  Agent Runtime ───────┐                                    │
│     │                  │ Incremental Changes                │
│     ▼                  ▼                                    │
│  ┌──────────────┐     ┌──────────────┐                     │
│  │ Daily Full    │◄───┼──► Backup Log │                     │
│  │ Backup        │    │ (PostgreSQL)  │                     │
│  └──────────────┘     └──────────────┘                     │
│                        │                                    │
│                        ▼                                    │
│              ┌─────────────────────────────┐                │
│              │ Encrypted S3 Storage        │◄── Retention ───│
│              │ (AWS/GCP/Azure)             │   (7-365 days)  │
│              └─────────────────────────────┘                │
└─────────────────────────────────────────────────────────────┘

Backup Types:

• Daily Full (2:00 AM WIB) - 100% state capture
• Incremental (5min intervals) - Changes only
• Manual (moltghost backup create)

Backup Lifecycle

graph TD
    A[Agent Running] --> B[Daily 2AM Full Backup]
    A --> C[5min Incremental]
    C --> D[Compress + Encrypt]
    D --> E[S3 Object Storage]
    
    F[Manual Backup] --> D
    
    G[Restore Request] --> H[Select Point-in-Time]
    H --> I[Rehydrate Pod]
    I --> J[Load Full + Incrementals]
    J --> K[Agent Identical State]
    
    style K fill:#90EE90

Retention Policies:

Free: 7 days (3 full backups)
Pro: 30 days (30 full backups) 
Enterprise: 365 days + custom

Restore Process

Zero-Downtime Recovery in 60-300 seconds:

# List available backups
moltghost backup list my-agent
# 2026-03-05T02:00:00Z  [FULL]  2.3MB
# 2026-03-04T14:30:00Z  [MANUAL] 1.8MB

# Restore to new/existing agent
moltghost restore my-agent-backup --target my-agent-restored

# Or point-in-time recovery
moltghost restore my-agent --timestamp 2026-03-04T14:30:00Z

Restore Flow:

1. Provision identical Pod spec
2. Download full + incrementals  
3. Decrypt + decompress (AES-256)
4. Rehydrate OpenClaw memory
5. Resume skills + configuration
6. Health check → Endpoint active ✅

Use Cases

Real Example:

Dev accidentally deleted production agent with 10K conversation memory
→ moltghost restore prod-agent --timestamp yesterday → Back online in 2min
Memory + skills perfectly preserved

Data Integrity & Security

┌─────────────────────────────────────────────────────────────┐
│                    Enterprise Grade                          │
├─────────────────────────────────────────────────────────────┤
│  ✅ AES-256 encryption at rest                              │
│  ✅ Customer-managed KMS keys (BYOK)                        │
│  ✅ CRC32 + SHA256 integrity checks                         │
│  ✅ Immutable backups (90-day WORM)                         │
│  ✅ Geo-redundant storage (3 copies)                        │
│  ✅ Point-in-time recovery (granular to 5min)               │
└─────────────────────────────────────────────────────────────┘

Compliance: SOC2, GDPR, ISO 27001 ready.

Cost Structure

Monthly Example (10 agents):

10GB backups × $0.03 = $0.30/month
Negligible vs compute savings from recovery

Summary

Backup & Restore = Agent Immortality

✅ Automated daily + incremental backups
✅ Point-in-time recovery (5min granularity)
✅ Enterprise encryption + integrity
✅ 60s recovery SLA for critical agents
✅ Cross-region disaster recovery

Never lose agent intelligence—restore any agent to any prior state instantly.

Next: Monitoring & Observability → Metrics, logs, alerts

Pro Tip: Enable --auto-backup=daily + --retention=90d for production agents.