openclaw-security-hardening

Run `npx skills add shanggqm/openclaw-security-hardening --all -g -y` in your terminal to install both security skills globally.

Just say 'security check' or '安全扫描' to your agent, and it audits your openclaw.json across seven dimensions like DM policy, gateway exposure, and file permissions, then gives you a scored report.

Ask your agent to install the skill, and the safe-install skill downloads it to a temp folder, runs a 7-point security review, and only installs it if no dangerous behavior is found.

You can pick from three levels: Basic fixes critical issues in about 5 minutes, Standard adds group chat and sandbox protection, and Strict applies maximum lockdown for public-facing deployments.

No. Both skills are prompt-only and work entirely through the agent's built-in read, edit, and exec tools, so there is nothing extra to install or maintain.