Claude Skill
ttguy0707/CyberClaw
CyberClaw is a transparent agent architecture with full behavior auditing, two-phase secure invocation, dual-level memory, and heartbeat tasks. Achieves 80% reduction in P0 incidents. Compatible wi...
Overview
Repository
Install this Skill
git clone https://github.com/ttguy0707/CyberClaw.gitRegistry
git clone https://github.com/ttguy0707/CyberClaw.gitpip install -e .git clone https://github.com/.../skill-creator.gitgit clone https://github.com/.../skill-vetter.gitpip install -e ".[dev]"
Summary
CyberClaw is a next-generation transparent agent architecture designed for enterprise AI safety and observability. It features full behavior auditing, two-phase secure invocation, dual-level memory, and heartbeat tasks, achieving an 80% reduction in P0 incidents. Compatible with OpenClaw and Claude Code skill ecosystem.
👾 下一代透明智能体架构 | Next-Gen Transparent Agent Architecture 🔍 全行为审计 | 🛡️ 两段式安全调用 | 🧠 双水位记忆 | ⏰ 心跳任务 📊 P0 级事故率降低 80% | 兼容 OpenClaw + Claude Code 技能生态
Key features
- Full behavior auditing for complete transparency
- Two-phase secure invocation for enhanced safety
- Dual-level memory (short-term & long-term)
- Heartbeat task mechanism for reliability
- 80% reduction in P0 incidents
- Compatible with OpenClaw and Claude Code skill ecosystem
Use cases
- Enterprise AI agent deployment with strict audit requirements
- Building safe and transparent autonomous agents
- Integrating with Claude Code for enhanced development workflows
- Cross-platform agent orchestration and monitoring
- Reducing critical incidents in production AI systems
README excerpt
<div align="center">  # CyberClaw ### **当 AI 开始"黑箱操作"，你需要一双透视眼** [](https://github.com/ttguy0707/CyberClaw) [](https://python.org) [](https://langchain-ai.github.io/langgraph/) [](https://python.langchain.com/) [](LICENSE) [](tests/) [](https://github.com/ttguy0707) **下一代透明智能体架构** · Next-Gen Transparent Agent Architecture [快速开始](#-快速开始) · [核心能力](#-核心能力) · [架构图](#-系统架构) · [示例](#-基本用法) </div> --- > 🤖 **你的 AI 在背着你做什么？CyberClaw 让所有行为无所遁形** > > 💡 **灵感来源**：受 [OpenClaw](https://github.com/openclaw/openclaw) 的启发，CyberClaw 专注于解决 AI 智能体的透明度和可控性问题。 --- ## 📖 简介 CyberClaw 是一个**企业级透明可控智能体**，重新定义 AI 系统的可信边界： - **🔍 白盒化决策** → 5 类事件审计 + JSONL 日志 + Rich 监控终端，所有行为可追溯 - **🛡️ 零信任执行** → 两段式调用（help → run），先看说明书再执行，P0 级事故率降低 80% - **🧠 持续学习** → 双水位记忆系统（长期画像 + 短期摘要），越用越懂你 - **⚡ 复杂任务编排** → 心跳任务系统 + 可插拔技能 + MCP 服务集成，解放双手 ### 🔌 技能生态兼容 CyberClaw 支持**OpenClaw 技能**和**Claude Code 技能**，可直接使用两个生态系统的丰富技能资源，无需重新开发。 ### 🌟 核心能力 | 能力 | 说明 | 优势 | |------|------|------| | **🧠 双水位记忆** | 长期画像 + 短期摘要，持续学习用户偏好 | 越用越懂你，避免重复询问 | | **🔍 全行为审计** | 5 类事件实时审计，JSONL 日志 + Rich 监控终端 | 告别黑箱，所有决策可追溯 | | **🛡️ 零信任执行** | help → run 两段式调用，先看说明书再执行 | P0 级事故率降低 80%（50% → 10%）| | **⏰ 心跳任务引擎** | 后台独立进程，自动执行定时任务 | 解放双手，复杂任务自动化 | | **🖥️ 跨平台支持** | Unix + Windows 双平台自适应，LLM 自主选择命令 | 一套代码，全平台运行 | --- ## ✨ 功能