{"skill":{"slug":"merlin-security-sentinel","displayName":"merlin-security-sentinel","summary":"Use this skill when the user asks about securing their OpenClaw installation, configuring AI agents safely, understanding prompt injection risks, dealing wit...","tags":{"agent-safety":"1.0.0","latest":"1.0.0","prompt-injection":"1.0.0","security":"1.0.0"},"stats":{"comments":0,"downloads":127,"installsAllTime":0,"installsCurrent":0,"stars":0,"versions":1},"createdAt":1774667666186,"updatedAt":1774668710760},"latestVersion":{"version":"1.0.0","createdAt":1774667666186,"changelog":"- Initial release of merlin-security-sentinel, providing guidance on securing OpenClaw installations and configuring AI agents safely.  \n- Documents risks associated with persistent AI agents: credential accumulation, memory poisoning, and supply chain attacks.  \n- Offers immediate hardening steps, including file permissions, restricted tool profiles, binding to localhost, allowlists, and skill audits.  \n- Emphasizes ephemeral execution for privileged tasks and explains architectural principles for agentic security.  \n- Lists practical defenses against prompt injection and clarifies the limitations of OpenClaw's security model.  \n- Includes quick reference answers to common OpenClaw security questions and incident responses.","license":"MIT-0"},"metadata":null,"owner":{"handle":"thepoorsatitagain","userId":"s17e3sa250vymf9qwb9bwmcxs583s46j","displayName":"thepoorsatitagain","image":"https://avatars.githubusercontent.com/u/253227633?v=4"},"moderation":null}