youseefhamdi/SPECTRA-CyberDefense
GitHub: youseefhamdi/SPECTRA-CyberDefense
一个融合了 LLM 自适应欺骗、RAG 威胁情报、ML 分类与 eBPF 封控技术的自主多层网络主动防御架构。
Stars: 1 | Forks: 0
# SPECTRA:自主主动网络防御与欺骗架构
[](https://opensource.org/licenses/MIT)
[](https://www.python.org/downloads/)
[](https://www.docker.com/)
**SPECTRA**(智能主动交战、反情报与威胁响应架构)是一种自主网络防御架构,集成了蜜罐编排、RAG 驱动的威胁情报、机器学习分类、自适应欺骗、内核级 eBPF/XDP 执行以及拜占庭容错多智能体编排。
## 🏗️ 架构概览
SPECTRA 跨越 6 个集成层,包含 14 个协议:
- **Layer 1(基础设施)**:P1 Hydra, P2 Don
- **Layer 2(检测)**:P3 Chrollo, P4 Hisoka
- **Layer 3(情报)**:P11 Meruem
- **Layer 4(作战)**:P5 Zetsu, P6 Knov, P7 Leorio
- **Layer 5(封控)**:P8 Seirin, P12 Netero, P13 Pitou, P14 Ging
- **Layer 6(编排)**:P9 Kurapika, P10 Zeno
## 📦 仓库结构
```
SPECTRA-CyberDefense/
├── protocols/
│ ├── p1_hydra/ # Clone orchestration (20 Docker containers)
│ ├── p2_don/ # RAG threat intelligence (Hybrid Dense/Sparse)
│ ├── p3_chrollo/ # ML classification (150 features, 94.6% accuracy)
│ ├── p4_hisoka/ # Adaptive LLM deception & attacker profiling
│ ├── p5_zetsu/ # eBPF process hiding & stealth (C Prototype)
│ ├── p6_knov/ # Multi-channel covert exfiltration
│ ├── p7_leorio/ # Self-healing health monitoring
│ ├── p8_seirin/ # XDP packet filtering & containment (C Prototype)
│ └── p10_zeno/ # PBFT multi-agent orchestration
├── docker/
│ └── honeypot-base/ # Base honeypot Dockerfile
├── requirements.txt # Python dependencies
└── README.md
```
## 🚀 快速开始
### 前置条件
- Python 3.10+
- Docker 24.0+
- 最低 8GB RAM
### 安装
```
git clone [https://github.com/youseefhamdi/SPECTRA-CyberDefense.git](https://github.com/youseefhamdi/SPECTRA-CyberDefense.git)
cd SPECTRA-CyberDefense
python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt
# 构建 honeypot base
cd docker/honeypot-base
docker build -t honeypot-base:latest .
```
### 使用示例
#### P1: Hydra(克隆编排)
```
from protocols.p1_hydra.hydra_controller import HydraController
hydra = HydraController(clone_count=5)
hydra.spawn_clones()
status = hydra.health_check()
```
#### P10: Zeno(PBFT 编排)
```
from protocols.p10_zeno.zeno_orchestrator import ZenoOrchestrator
zeno = ZenoOrchestrator()
threat = {"classification": "malicious", "attacker_ip": "192.168.1.100"}
result = zeno.orchestrate_response(threat)
```
## 📊 性能指标
| 协议 | 指标 | 数值 |
|----------|--------|-------|
| P1 Hydra | Clone spawn time | 2.3s |
| P2 Don | Retrieval latency (p95) | 47ms |
| P3 Chrollo | Classification accuracy | 94.6% |
| P10 Zeno | Consensus latency (p95) | 89ms |
## 📝 引用
```
@article{youssefhamdi2026spectra,
title={SPECTRA: Autonomous Active Cyber Defense Architecture},
author={youssef hamdi zafaan ibrahim , mohammed khalaf salama},
year={2026}
}
```
## 📄 许可证
MIT 许可证 - 详情请参阅 LICENSE 文件。
## 👥 作者
- **Youssef Hamdi Zafaan Ibrahim** - 电信与电子理学学士最后一年
Bug Hunter & Red Teamer。
- **Dr. Mohammed Khalaf Salama** - 研究导师
**⚠️ 免责声明**:仅供教育和防御目的的研究原型。
标签:Apex, BOF, C2, CTF学习, DLL 劫持, Docker, Docker镜像, GitHub, Google搜索, IP 地址批量处理, LLM, ML分类, PBFT, Python, RAG, Unmanaged PE, XDP, 内核级安全, 内网渗透, 域名收集, 多智能体, 大语言模型, 威胁情报, 安全防御评估, 实时处理, 容器编排, 密码管理, 开发者工具, 恶意流量检测, 拜占庭容错, 攻击画像, 数据渗漏, 无后门, 机器学习, 检索增强生成, 欺骗防御, 流量过滤, 网络安全, 自主防御, 自愈系统, 蜜罐, 证书利用, 请求拦截, 逆向工具, 隐私保护, 隐蔽信道