56个即用技能，使你的AI代理具备自主、自我修复和自我提升的能力。

OpenClaw 即插即用的技能库——开源的 AI 代理运行时。赋予你的代理结构化思维、安全防护、持久内存、cron 调度、自我恢复，以及在对话中编写新技能的能力。

为那些希望AI代理全天候24小时自动运行，而不仅仅是在聊天窗口中回复提示的开发者打造。

灵感来自奥布拉/超能力。扩展为永不眠的特工。

---

为什么会有这样的存在

大多数AI代理框架给你的是一个聊天机器人，会在会话间忘记所有内容。OpenClaw则不同——它持续运行，处理数小时任务，并有原生的cron调度功能。但开箱即用时，它并不知道如何好好利用这些功能。

openclaw-superpowers弥合了这一差距。 在一个命令中安装56项技能，你的代理立刻知道如何：

• 先思考再行动 — 头脑风暴、规划和系统调试技能防止“冲进去破坏”的失败模式

openclaw-superpowers

56 ready-to-use skills that make your AI agent autonomous, self-healing, and self-improving.

A plug-and-play skill library for OpenClaw — the open-source AI agent runtime. Gives your agent structured thinking, security guardrails, persistent memory, cron scheduling, self-recovery, and the ability to write its own new skills during conversation.

Built for developers who want their AI agent to run autonomously 24/7, not just respond to prompts in a chat window.

Inspired by obra/superpowers. Extended for agents that never sleep.

---

Why this exists

Most AI agent frameworks give you a chatbot that forgets everything between sessions. OpenClaw is different — it runs persistently, handles multi-hour tasks, and has native cron scheduling. But out of the box, it doesn’t know how to use those capabilities well.

openclaw-superpowers bridges that gap. Install 56 skills in one command, and your agent immediately knows how to:

• Think before it acts — brainstorming, planning, and systematic debugging skills prevent the “dive in and break things” failure mode
• Protect itself — 6 security skills detect prompt injection, block dangerous actions, audit installed code, and scan for leaked credentials
• Run unattended — 12 cron-scheduled skills handle memory cleanup, health checks, budget tracking, and community monitoring while you sleep
• Recover from failures — self-recovery, loop-breaking, and task handoff skills keep long-running work alive across crashes and restarts
• Never forget — DAG-based memory compaction, integrity checking, context scoring, and SQLite session persistence ensure the agent preserves critical information even in month-long conversations
• Improve itself — the agent can write new skills during normal conversation using create-skill, encoding your preferences as permanent behaviors

---

The self-modifying agent

This is what makes openclaw-superpowers different from every other plugin library:

“Every time I do a code review, check for security issues first.”

Your agent invokes create-skill, writes a new SKILL.md, and that behavior is live — immediately, permanently, no restart needed. The agent encodes your preferences as durable skills. You describe what you want. It teaches itself.