OpenClaw Secure Baseline

A practical security baseline for running messaging automation safely on personal devices, home labs, and small business environments.

What This Solves

Messaging bots connected to platforms like iMessage, WhatsApp, Telegram, and Signal face real risks:

• Spam-triggered automation — unknown senders causing unintended actions
• Bot presence detection — auto-replies revealing your bot exists
• Enumeration attacks — different error messages leaking sender status
• Accidental exposure — control gateways reachable from LAN/internet
• Weak defaults — insecure authentication left enabled

This baseline prevents all of these.

Core Principles

Principle	Implementation
Default Deny	Unknown senders are blocked, not queued
Silent Drop	No reply to unknown senders (prevents oracle)
Owner-Initiated Pairing	Codes generated only when owner acts
Least Privilege	Approved senders get minimal capabilities
Defense in Depth	Multiple independent security layers

Components

Message Gatekeeper

Zero-trust sender validation with:

• Sender registry (approved/pending/blocked)
• Owner-initiated pairing flow
• Capability-based access control (RBAC)
• Privacy-safe audit logging (HMAC hashes)

Gateway Security Check

Regression guard that verifies:

• Gateway bound to localhost only
• Insecure auth flags disabled
• No Docker/proxy exposure

Quick Start

# 1. Copy the gatekeeper to your project
cp src/message_gatekeeper.py /your/project/security/

# 2. Set up the CLI wrapper
cp scripts/gatekeeper /your/project/security/
chmod +x /your/project/security/gatekeeper

# 3. Configure your gateway (see examples/)
# Ensure bind = "loopback" and insecure flags = false

# 4. Schedule the security check
cp scripts/gateway-security-check.sh /your/project/security/
chmod +x /your/project/security/gateway-security-check.sh

Documentation

• SECURITY.md — Security policy and principles
• docs/security/HARDENING.md — Step-by-step hardening guide

What This Is NOT

• Not a vulnerability disclosure
• Not an exploit repository
• Not enterprise compliance documentation
• Not a penetration testing guide

Scope & Assumptions

This baseline assumes:

• Host OS is trusted
• Admin credentials are not compromised
• Physical access is controlled

Threats outside this scope are intentionally excluded.

Philosophy

Automation should be secured like an API, not treated like a chat UI.

License

MIT License — see LICENSE