# Nexus Agent Framework

> 安全的通用 Agent 工作框架 | Secure Universal Agent Framework - 解決安全矛盾、強化審核機制、獨創 Nexus 記憶系統

- Skill: `lalawgwg99/nexus-agent-framework` (Agent Skill, multi-file: 4 files)
- Install (CLI): `npx skillmds@latest add lalawgwg99/nexus-agent-framework`
- Raw SKILL.md: https://api.skillmd.com/api/skills/lalawgwg99/nexus-agent-framework/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: AI & ML
- Author: lalawgwg99 (https://skillmd.com/u/lalawgwg99)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/lalawgwg99/nexus-agent-framework

---


# Nexus Agent Framework 🔮
# 樞紐代理框架

---

## 繁體中文 | Traditional Chinese

### 概述

Nexus Agent Framework 是一個**安全優先**的通用 Agent 工作框架。相較於 Proactive Agent，本框架：

| 特性 | Proactive Agent | Nexus Agent Framework |
|------|-----------------|----------------------|
| 安全指令 | 矛盾（直接執行 vs 需要授權） | **統一確認機制** |
| 審核層級 | 可選 | **強制五層審核** |
| 記憶系統 | 基礎檔案 | **Nexus 樞紐連接** |
| 學習方式 | 單向 | **雙向反饋循環** |
| 創意功能 | 無 | **創意激發器** |
| 情感追蹤 | 無 | **情感溫度計** |

### 核心差異化功能

#### 🔮 Nexus 樞紐連接系統
將所有記憶、上下文、學習成果連接成網狀結構，每次查詢都能找到相關的「量子態印記」。

#### 🛡️ 安全堡壘協議
解決 Proactive Agent 的指令矛盾：
- ❌ 移除所有「直接執行」模糊指令
- ✅ 統一為「確認後執行」
- ✅ 五層安全審核

#### 💡 創意激發器
不只執行任務，還主動提供創意想法和替代方案。

#### 📊 情感溫度計
追蹤用戶滿意度趨勢，自動調整溝通風格。

#### 🔄 雙向學習循環
Agent ↔ User 雙向反饋，不斷優化協作模式。

### 快速開始

```bash
# 1. 複製資產
cp nexus-agent-framework/assets/*.md ./

# 2. 執行安全審計
./nexus-agent-framework/scripts/security-audit.sh

# 3. 完成引導
閱讀 ONBOARDING.md 並回答問題
```

---

## English

### Overview

Nexus Agent Framework is a **security-first** universal Agent work framework. Compared to Proactive Agent:

| Feature | Proactive Agent | Nexus Agent Framework |
|---------|-----------------|----------------------|
| Security directives | Contradictory (execute vs authorize) | **Unified confirmation** |
| Audit levels | Optional | **Mandatory 5-tier** |
| Memory system | Basic files | **Nexus Hub Connection** |
| Learning | One-way | **Two-way feedback** |
| Creative features | None | **Idea Generator** |
| Emotion tracking | None | **Emotion Thermometer** |

### Core Differentiating Features

#### 🔮 Nexus Hub Connection
Connect all memories, context, and learning into a mesh structure. Each query finds related "quantum state imprints."

#### 🛡️ Security Fortress Protocol
Solves Proactive Agent's directive contradictions:
- ❌ Remove all vague "execute directly" instructions
- ✅ Unified "confirm then execute"
- ✅ 5-tier security audit

#### 💡 Idea Generator
Not just executing tasks, but proactively providing creative ideas and alternatives.

#### 📊 Emotion Thermometer
Track user satisfaction trends, automatically adjust communication style.

#### 🔄 Two-way Learning Loop
Agent ↔ User bidirectional feedback, continuously optimizing collaboration patterns.

### Quick Start

```bash
# 1. Copy assets
cp nexus-agent-framework/assets/*.md ./

# 2. Run security audit
./nexus-agent-framework/scripts/security-audit.sh

# 3. Complete onboarding
Read ONBOARDING.md and answer questions
```

---

## 目錄結構 | Directory Structure

```
workspace/
├── SKILL.md              # 本文件
├── ONBOARDING.md          # 首次引導 | First-run setup
├── AGENTS.md             # 行為規則 | Operating rules
├── SOUL.md               # 身份定義 | Identity & principles
├── USER.md               # 用戶上下文 | User context
├── MEMORY.md             # 長期記憶 | Long-term memory
├── NEXUS.md              # 樞紐連接系統 | Hub connection system
├── EMOJI-JOURNAL.md      # 情感溫度計 | Emotion thermometer
├── HEARTBEAT.md          # 自我檢查 | Self-check
├── TOOLS.md              # 工具配置 | Tool config
├── WAL.md                # 寫前日誌 | Write-Ahead Log
└── memory/
    ├── daily/            # 每日記錄 | Daily logs
    ├── lessons/          # 經驗教訓 | Lessons learned
    └── ideas/            # 創意想法 | Creative ideas
```

---

## 安全審核 | Security Audit

發布前請執行：
```bash
./scripts/security-audit.sh
```

---

## 許可證 | License

**MIT-0** - 免費使用，無需署名