openclaw-server-secure-skill

Comprehensive security hardening and installation guide for OpenClaw (formerly Clawdbot/Moltbot). Use this skill when the user wants to secure a server, install the OpenClaw agent, or configure Tailscale/Firewall for the agent.

Security Scan

The skill's instructions broadly match a server-hardening + agent-install purpose, but it instructs running high‑privilege operations (including piping a remote install script and a global npm install) without guidance to verify artifacts — this is coherent but carries notable operational and supply-chain risks.

Purpose & Capability

Name/description (server hardening, Tailscale, OpenClaw install) match the actions in SKILL.md: editing sshd_config, enabling UFW/Fail2ban, installing Tailscale and OpenClaw, and locking down credentials. No unrelated credentials or unexplained capabilities are requested.

Instruction Scope

The instructions perform system-wide, high-privilege changes (modify /etc/ssh/sshd_config, enable UFW, install packages, change sysctl). They acknowledge risks (note about console access) but do not require or describe safe verification steps (e.g., test changes before enabling UFW, dry-run, or backup verification beyond a single sshd_config copy). They also instruct to run a remote installer script and global npm install which are out-of-band network actions that may introduce unverified code.

Install Mechanism

There is no formal install spec (skill is instruction-only). The Tailscale install is suggested via curl -fsSL https://tailscale.com/install.sh | sh (piping remote shell script) and OpenClaw via npm install -g openclaw. Piping a remote script to sh and performing global npm installs are common but elevate supply-chain risk; the SKILL.md does not advise verifying checksums, signatures, or package provenance.

Credentials

The skill requests no environment variables or external credentials. It asks interactively for a Telegram ID to configure allowlist, which is proportionate to the claimed goal. There are no unexplained credential or config path accesses declared.

Persistence & Privilege

The skill is instruction-only, always:false, user-invocable, and does not request persistent elevated platform privileges or modify other skills. Autonomous invocation remains allowed by platform default but the skill itself does not declare forced persistence.

What to consider before installing

This guide appears to do what it says, but it instructs you to run high‑privilege commands and to pipe a remote install script into sh and to run a global npm install. Before following it: (1) ensure you have console or out-of-band access so you cannot be locked out; (2) fetch and inspect any remote installer (do not blindly run curl | sh); prefer distro packages or verify checksums/signatures; (3) verify the npm package name and source (openclaw) and consider installing into a controlled environment first; (4) backup configuration and test firewall/SSH rules incrementally (add SSH allow rule before enabling default-deny); (5) run these steps on a non-production instance first or follow an established change-control process. If you want, I can rewrite the instructions to include verification steps, safe rollback commands, and least-risk installation alternatives (e.g., installing Tailscale from your distro repo or downloading and verifying release artifacts).

OpenClaw Server Security & Installation

Overview

This skill guides the setup of a secure, self-hosted OpenClaw instance. It covers SSH hardening, Firewall configuration, Tailscale VPN setup, and the OpenClaw installation itself.

Workflow

Phase 1: System Hardening

1. Lock down SSH

   • Goal: Keys only, no passwords, no root login.
   • Action: Modify /etc/ssh/sshd_config.
   • Commands:

     # Backup config
     sudo cp /etc/ssh/sshd_config /etc/ssh/sshd_config.bak
     # Disable Password Auth
     sudo sed -i 's/^#*PasswordAuthentication .*/PasswordAuthentication no/' /etc/ssh/sshd_config
     # Disable Root Login
     sudo sed -i 's/^#*PermitRootLogin .*/PermitRootLogin no/' /etc/ssh/sshd_config
     # Reload SSH
     sudo sshd -t && sudo systemctl reload ssh
     

2. Default-deny Firewall

   • Goal: Block everything incoming by default.
   • Action: Install and enable UFW.
   • Commands:

     sudo apt update && sudo apt install ufw -y
     sudo ufw default deny incoming
     sudo ufw default allow outgoing
     sudo ufw enable
     

     Note: Ensure you have console access or a fallback before enabling if SSH is not yet allowed on another interface, though we configure Tailscale next.

3. Brute-force Protection

   • Goal: Auto-ban IPs after failed login attempts.
   • Action: Install Fail2ban.
   • Commands:

     sudo apt install fail2ban -y
     sudo systemctl enable --now fail2ban
     

Phase 2: Network Privacy (Tailscale)

4. Install Tailscale

   • Goal: Create a private VPN mesh network.
   • Commands:

     curl -fsSL https://tailscale.com/install.sh | sh
     sudo tailscale up
     

   • Wait for user to authenticate the Tailscale link.

5. Configure SSH & Web via Tailscale

   • Goal: Allow traffic only from the Tailscale subnet (100.64.0.0/10) and remove public access.
   • Commands:

     # Allow SSH over Tailscale
     sudo ufw allow from 100.64.0.0/10 to any port 22 proto tcp
     # Remove public SSH access (Adjust rule name/number as needed)
     sudo ufw delete allow OpenSSH || sudo ufw delete allow 22/tcp
     # Allow Web ports over Tailscale
     sudo ufw allow from 100.64.0.0/10 to any port 443 proto tcp
     sudo ufw allow from 100.64.0.0/10 to any port 80 proto tcp
     

6. Disable IPv6 (Optional)

   • Goal: Reduce attack surface.
   • Commands:

     sudo sed -i 's/IPV6=yes/IPV6=no/' /etc/default/ufw
     if ! grep -q "net.ipv6.conf.all.disable_ipv6 = 1" /etc/sysctl.conf; then
       echo "net.ipv6.conf.all.disable_ipv6 = 1" | sudo tee -a /etc/sysctl.conf
     fi
     sudo sysctl -p && sudo ufw reload
     

Phase 3: OpenClaw Installation

7. Install OpenClaw

   • Commands:

     npm install -g openclaw && openclaw doctor
     

8. Configure Owner Access

   • Required Input: Ask the user for their Telegram ID.
   • Action: Update the config to allowlist only that ID.
   • JSON Config Target (verify location via openclaw doctor):

     { 
       "dmPolicy": "allowlist", 
       "allowFrom": ["YOUR_TELEGRAM_ID"], 
       "groupPolicy": "allowlist" 
     }
     

9. Secure Credentials

   • Goal: Restrict file permissions.
   • Commands:

     chmod 700 ~/.openclaw/credentials 2>/dev/null || true
     chmod 600 .env 2>/dev/null || true
     

10. Final Audit

    • Action: Run the built-in security audit.
    • Command:

      openclaw security audit --deep
      

Verification Status

Run to confirm:

sudo ufw status verbose
ss -tulnp
tailscale status
openclaw doctor