moltbot-security
por nextfrontierbuilds
MoltbookResumo
Security hardening guide for AI agents.
Descrição
O que esta skill faz
A configuration and hardening guide for OpenClaw and Moltbot AI agent gateways. Addresses publicly exposed gateways, which can leak API keys, conversation histories, and shell access to any internet scanner. Based on research that found 1,673+ exposed instances on Shodan.
It consolidates the most exploited gateway misconfigurations into a single audit command and config template, replacing ad-hoc research when deploying AI agents on cloud servers.
Quando usá-la
- Locking down an OpenClaw gateway before exposing it beyond localhost
- Auditing an existing Moltbot setup for common misconfigurations
- Setting up Tailscale to access your AI agent remotely without opening a public port
- Hardening a cloud server running OpenClaw with UFW firewall rules
- Preventing prompt injection attacks that exfiltrate private conversation histories
Exemplo de fluxo de trabalho
User asks: audit my OpenClaw gateway and fix any security issues
- 1
Runs
openclaw security audit --deepto detect binding, auth, and permission issues - 2
Identifies gateway bound to all interfaces instead of loopback
- 3
Applies
openclaw security audit --deep --fixto correct file permissions and binding - 4
Generates a secure token with
openssl rand -hex 32and writes it to the config - 5
Verifies the updated config matches the secure defaults template
Gateway rebound to loopback, auth token set, file permissions corrected to 600/700, mDNS broadcasting disabled
Requisitos
- Tailscale account (only required if using remote access via Tailscale)
Instalação
clawhub install nextfrontierbuilds/moltbot-securityExecute o comando da CLI ou cole o prompt no Claude Code, Cursor, Codex ou no agente de código que você usa.