🦀 OpenClaw Security Scan

Production-grade security audits for OpenClaw AI agent runtimes.

workflow

graph LR
    A[Start Scan] --> B{Discover Configs}
    B -->|Found| C[Parse Configs]
    B -->|None| D[System Checks Only]
    C --> E[Run 12 Security Checks]
    D --> E
    E --> F[Generate Report]
    F --> G{Critical Issues?}
    G -->|Yes| H[Fail Build / Suggest Fix]
    G -->|No| I[Pass]

Loading

Features

• Zero Dependencies: Written in pure Bash (runs on Linux, macOS, WSL).
• Auto-Fix: Automatically remediates critical issues like exposed ports and missing allowlists.
• CI/CD Ready: Outputs JSON reports for pipeline integration.
• Smart Discovery: Automatically finds configurations in .openclaw, .moltbot, .clawdbot, and environment variables.

Install

Quick scan (no install):

curl -sL https://raw.githubusercontent.com/legendaryabhi/openclaw-security-scan/main/scan.sh | bash

Install locally (recommended — enables fix command):

curl -sL https://raw.githubusercontent.com/legendaryabhi/openclaw-security-scan/main/scan.sh -o scan.sh
chmod +x scan.sh
./scan.sh scan

Usage

# Full security audit
./scan.sh scan

# JSON output (for CI/CD pipelines)
./scan.sh scan --json

# Auto-fix critical issues (interactive)
./scan.sh fix

# Auto-fix without confirmation (for scripts)
./scan.sh fix --yes

Security Checks (12)

#	Check	Severity	What it detects
1	Gateway Binding	CRITICAL	Gateway listening on 0.0.0.0 (exposed to network) instead of 127.0.0.1.
2	API Keys Exposure	CRITICAL	Plaintext API keys (OpenAI, Anthropic, etc.) hardcoded in config files instead of .env.
3	File Permissions	CRITICAL	Config files readable by other users (should be 600).
4	HTTPS/TLS	WARNING	No TLS/HTTPS configuration detected for the gateway.
5	Shell Command Allowlist	CRITICAL	Missing tools.exec.safeBins — allows agent to run arbitrary system commands.
6	Sensitive Directories	WARNING	Agent has access to sensitive paths like ~/.ssh, ~/.gnupg, /etc/shadow.
7	Webhook Auth	WARNING	Webhook endpoints configured without authentication tokens.
8	Sandbox Isolation	WARNING	Agent not running inside a Docker container or sandbox environment.
9	Default/Weak Credentials	CRITICAL	Use of default tokens (changeme, admin) or undefined token bug.
10	Rate Limiting	WARNING	No rate limiting configured, leaving the agent open to DoS.
11	Node.js Version	CRITICAL	Vulnerable Node.js versions susceptible to CVE-2026-21636 (permission model bypass).
12	Control UI Auth	CRITICAL	Auth bypass flag enabled for the Control UI.

Fix Command

The fix command (./scan.sh fix) automatically remediates the following critical issues:

1. Rebinds Gateway: Changes host: 0.0.0.0 to 127.0.0.1.
2. Fixes Permissions: Sets config file permissions to 600 (read/write by owner only).
3. Adds Allowlist: Injects a safe default safeBins list (ls, grep, cat, etc.) if missing.
4. Enables Auth: Disables authBypass flags for the Control UI.

Note: A timestamped backup of your config is created in ~/.scan-backup/ before any changes are made.

Contributing

Contributions are welcome!

Read the CONTRIBUTING.md for details.

License

MIT — see LICENSE.

---

Maintained by Abhinav Sharma