OpenClaw·Docker·Docker Compose·VPS·UFW
⚠️ Critical Warning: Docker Is Not Enough
Containerization is a layer, not a solution.
Running OpenClaw in Docker without additional isolation leaves you vulnerable to:
- Container escape via privileged mode or kernel exploits
- Host network exposure if the gateway binds to 0.0.0.0
- Volume mount risks when sharing host filesystems
- Secret leakage through environment variables or mounted files
You need defense in depth. This guide gives you the Docker layer. Pair it with:
- Off-network VPS deployment (see YOLO Safely)
- Network segmentation and egress filtering
- Read-only filesystems and non-root execution
- Monitoring and audit logging
Coming soon: Dedicated VPS hardening guide with cloud-init scripts, firewall rules, and automated lockdown procedures (~Week 2-3 on content calendar).
Quick Start (If You Accept the Risks)
1
2
3
4
5
6
7
8
9
| # Clone the compose configuration
curl -fsSL https://raw.githubusercontent.com/openclaw/openclaw/main/docker-compose.yml -o docker-compose.yml
# Create environment file
cp .env.example .env
# Edit .env with your API keys and settings
# Start with security flags
docker-compose up -d
|
Security-Hardened Docker Compose
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
| version: '3.8'
services:
openclaw:
image: openclaw/openclaw:latest
container_name: openclaw-agent
restart: unless-stopped
# Security: Run as non-root
user: "1000:1000"
# Security: Read-only root filesystem
read_only: true
# Security: Limit capabilities
cap_drop:
- ALL
cap_add:
- NET_BIND_SERVICE # Only if binding to privileged port
# Security: No new privileges
security_opt:
- no-new-privileges:true
# Security: Limited resources
deploy:
resources:
limits:
cpus: '2.0'
memory: 2G
reservations:
cpus: '0.5'
memory: 512M
# Network: Custom isolated network
networks:
- openclaw-net
# Ports: Only expose what you need
# WARNING: Binding to 0.0.0.0 exposes to the internet
# Use 127.0.0.1:port:port for localhost-only
ports:
- "127.0.0.1:3000:3000" # Web UI - localhost only
- "127.0.0.1:8080:8080" # Gateway API - localhost only
# Environment: Pass through, not hardcoded
env_file:
- .env
# Volumes: Minimal, temporary where possible
volumes:
# Persistent data (encrypted at rest ideally)
- openclaw-data:/app/data
# Temporary files (noexec, nosuid, nodev)
- type: tmpfs
target: /tmp
tmpfs:
size: 100M
mode: 1777
# Read-only config
- ./config:/app/config:ro
# Health checks
healthcheck:
test: ["CMD", "curl", "-f", "http://localhost:8080/health"]
interval: 30s
timeout: 10s
retries: 3
start_period: 40s
# Logging
logging:
driver: "json-file"
options:
max-size: "10m"
max-file: "3"
labels: "openclaw,agent"
# Optional: Monitoring sidecar
agent-monitor:
image: prom/node-exporter:latest
container_name: openclaw-monitor
restart: unless-stopped
networks:
- openclaw-net
volumes:
- /proc:/host/proc:ro
- /sys:/host/sys:ro
- /:/rootfs:ro
command:
- '--path.procfs=/host/proc'
- '--path.rootfs=/rootfs'
- '--path.sysfs=/host/sys'
- '--collector.filesystem.mount-points-exclude=^/(sys|proc|dev|host|etc)($$|/)'
networks:
openclaw-net:
driver: bridge
internal: false # Set to true for no external access (use with proxy)
ipam:
config:
- subnet: 172.20.0.0/16
volumes:
openclaw-data:
driver: local
|
Environment Configuration
Create .env:
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
| # Required: API keys (use burner keys, not production)
ANTHROPIC_API_KEY=sk-ant-your-burner-key-here
OPENAI_API_KEY=sk-your-burner-key-here
# Optional: Model selection
DEFAULT_MODEL=claude-sonnet-4-5
FALLBACK_MODEL=kimi-k2-5
# Security: Disable dangerous features by default
ENABLE_SHELL_EXECUTION=false
ENABLE_FILE_WRITE=false
ENABLE_NETWORK_FETCH=false
# Only enable after reviewing skill code:
# ENABLE_SHELL_EXECUTION=true
# ENABLE_FILE_WRITE=true
# ENABLE_NETWORK_FETCH=true
# Logging
LOG_LEVEL=info
LOG_FORMAT=json
# Gateway binding (127.0.0.1 = localhost only)
GATEWAY_BIND=127.0.0.1
GATEWAY_PORT=8080
|
Network Security Checklist
Before Starting the Container
After Starting the Container
1
2
3
4
5
6
7
8
9
10
11
12
13
| # Verify network exposure
ss -tlnp | grep -E '(3000|8080)'
# Should show 127.0.0.1, NOT 0.0.0.0 or :::
# Check for unexpected ports
docker exec openclaw-agent ss -tlnp
# Verify read-only filesystem
docker exec openclaw-agent touch /tmp/test-write # Should work
docker exec openclaw-agent touch /app/test-write # Should fail
# Check running user
docker exec openclaw-agent id # Should show uid=1000, not root
|
Common Mistakes
Mistake 1: --privileged Mode
Don’t:
1
| docker run --privileged openclaw/openclaw # Gives full host access
|
Do: Use capability dropping (see compose file above).
Mistake 2: Volume Mounting Host Docker Socket
Don’t:
1
2
| volumes:
- /var/run/docker.sock:/var/run/docker.sock # Container escape
|
Do: Run separate monitoring containers, not Docker-in-Docker.
Mistake 3: Binding to 0.0.0.0
Don’t:
1
2
| ports:
- "3000:3000" # Binds to all interfaces
|
Do:
1
2
| ports:
- "127.0.0.1:3000:3000" # Localhost only
|
Mistake 4: Running on Your Laptop
Don’t: Run OpenClaw Docker on your development machine with access to:
- Your SSH keys (~/.ssh)
- Your Git config (~/.gitconfig)
- Your work Slack/Discord
- Your cloud provider credentials
Do: Deploy to a disposable VPS. See YOLO Safely.
Monitoring Container Activity
1
2
3
4
5
6
7
8
9
10
11
| # Real-time container stats
docker stats openclaw-agent
# View logs
docker logs -f openclaw-agent
# Inspect network connections
docker exec openclaw-agent netstat -tuln
# Check filesystem changes (run from host)
docker diff openclaw-agent
|
What’s Missing (Coming Soon)
This Docker configuration is one layer of a complete security strategy.
VPS Hardening Guide (~Week 2-3):
- Cloud-init scripts for automated lockdown
- UFW/iptables rules for egress filtering
- SSH hardening (key-only, port change, fail2ban)
- Host-level audit logging with auditd
- Automated snapshot/backup before experiments
- WireGuard mesh for distributed agent topologies
Subscribe to updates or check the content calendar for release timeline.
Last updated: February 3, 2026
Security verification: Docker configurations tested on Ubuntu 22.04 with Docker 24.0+. Always verify network binding before exposing to any network.