Security
Security is a core priority for AutoClaw. Here's how we protect your data and your AI assistant.
Infrastructure Security
Private Networking
Every AutoClaw instance runs on a private Tailscale network:
- Not exposed to the public internet — Your instance has no public IP
- End-to-end encryption — All traffic encrypted with WireGuard
- Zero-trust architecture — Every connection is authenticated
Isolated Instances
Your OpenClaw instance runs in isolation:
- Dedicated container per customer
- No shared resources with other users
- Network isolation between instances
Secure Hosting
Our infrastructure runs on enterprise-grade cloud providers:
- SOC 2 compliant data centers
- Regular security audits
- Automated patching and updates
Data Security
Encryption
| Data Type | At Rest | In Transit | |-----------|:-------:|:----------:| | Conversations | ✅ AES-256 | ✅ TLS 1.3 | | API Keys | ✅ AES-256 | ✅ TLS 1.3 | | Configuration | ✅ AES-256 | ✅ TLS 1.3 | | Logs | ✅ AES-256 | ✅ TLS 1.3 |
Credential Management
- API keys are encrypted at rest
- Keys are never logged or displayed in full
- Secrets are stored separately from application data
Data Retention
- Conversations are retained until you delete them
- Logs are retained for 30 days
- You can export or delete your data at any time
Authentication
Account Security
- Password hashing with bcrypt
- Optional two-factor authentication (2FA)
- OAuth support (Google, GitHub)
- Session management with secure tokens
Gateway Authentication
Your OpenClaw Gateway requires authentication:
- Gateway token generated during setup
- Tokens are rotated automatically
- Failed auth attempts are rate-limited
Skill Security
Vetted Skills Only
Every skill in the AutoClaw catalog is security-reviewed:
- Code auditing for malicious behavior
- API credential handling verification
- Sandboxed execution where possible
- Clear permission scopes
OAuth Best Practices
For OAuth-connected skills:
- Minimal permission scopes requested
- Tokens stored encrypted
- Automatic token refresh
- Revocation support
Sandboxing
Skills that execute code run in sandboxed environments:
- Limited file system access
- Network restrictions
- Resource limits (CPU, memory)
- Time limits
Agent Security
Tool Policies
Control what your agent can do:
- Elevated tools — Require explicit approval for dangerous operations
- Sandboxed execution — Code runs in restricted environments
- Approval workflows — Human-in-the-loop for sensitive actions
Rate Limiting
Protect against runaway agents:
- Message rate limits
- Token usage limits
- Tool call limits
Access Control
User Permissions
| Role | Capabilities | |------|--------------| | Owner | Full access, billing, delete instance | | Admin | Configuration, skills, channels | | User | Chat, view dashboard |
Audit Logging
All actions are logged:
- Who performed the action
- What action was taken
- When it occurred
- Success or failure
View audit logs in Settings → Security → Audit Log.
Incident Response
If a security issue occurs:
- Detection — Automated monitoring detects anomalies
- Containment — Affected resources are isolated
- Investigation — Our team investigates the cause
- Resolution — Issue is fixed and verified
- Communication — Affected users are notified
Reporting Security Issues
Found a security vulnerability? Contact us:
- Email: security@autoclaw.io
- We respond within 24 hours
- Responsible disclosure is appreciated
Compliance
AutoClaw is designed with compliance in mind:
- GDPR data handling
- Data export and deletion support
- Audit logging for accountability
Learn More
- OpenClaw Security — Gateway security details
- Sandboxing — How code execution is sandboxed