openclaw-runbook-security

Skill by ara.so — Hermes Skills collection.

Practical runbook for running OpenClaw AI agents day-to-day without burning money, exposing your gateway, or trusting unvetted automation. Based on the community-maintained openclaw-runbook project that provides opinionated, security-first guidance for production OpenClaw deployments.

What OpenClaw Is

OpenClaw is an open-source AI agent framework that runs locally or on a VPS, capable of autonomous task execution, memory management, and multi-model routing. It exposes a gateway for control and can spawn specialized sub-agents.

Key challenge: OpenClaw is powerful but can be expensive, insecure, or noisy without proper configuration.

This runbook's approach:

• Tailscale-first access (no public ports)
• Conservative model routing (cheap models for routine tasks)
• Manual skill vetting (inspect before install)
• Explicit memory boundaries
• Automation with guardrails