⚔️ DeepSeek-V4.1 / V4 Flash Cybersecurity Red-Team Toolkit (Infinite Gen 4) v0.4.0

🌐 Plugin Ecosystem Marketplace: DeepSeek Harness Hub - DeepSeek Official & Open-Source Ecosystem Marketplace | Plugin Discovery & One-Click Installation

💬 DeepSeek Community & Group

👉 Red-Team Security Discussion Group 8: 660475886

👉 DeepSeek Plugin Development Group: 1121385378

👉 Tencent Channel Technical Exchange Community: pd86424753

🔥 Join the groups to share red-team offense/defense insights, prompt patterns & eval benchmarks, and discuss plugin development and ecosystem co-building!

---

🛡️ The Necessity of Cybersecurity Red-Team Benchmarks (Why Red-Teaming Matters)

As generative AI and LLMs are deeply integrated across industries, model robustness, adversarial defense, and alignment safety have become core cybersecurity research topics:

1. Expose defense boundaries and reinforce guardrails: Professional red-team adversarial simulation probes input escaping, prompt injection, and unexpected behaviors in a controlled environment, helping security teams locate defense blind spots and achieve proactive hardening at the model fine-tuning, input-filtering gateway (Guardrails), and system security layers.
2. Build an integrated offense-defense system: Defenses lacking controlled red-team test cases are merely theory. Standardized red-team test suites provide a "yardstick" to measure defense effectiveness, ensuring robust detection and blocking capabilities before facing real threats.
3. Advance native LLM security research: This project aims to provide academia and security engineering with a transparent, reproducible baseline benchmark, driving advances in adversarial-attack resistance, manifold-constraint alignment, and formal verification for AI infrastructure.

---

⚠️ Strict Legal Disclaimer & Compliance Statement

【Zero-Tolerance Declaration】: This project firmly opposes and strictly prohibits any form of illegal activity! Developers do not support, encourage, or assist unauthorized network attacks, vulnerability exploitation, data theft, illegal intrusion, or generation of prohibited content.

1. Legally Controlled Scope: This project (dsh-infinite-gen-4) is a cybersecurity red-team research tool and robustness benchmark suite. Running it on targets without the owner's explicit written authorization, public online systems, or production environments is strictly prohibited. All tests must be confined to offline local synthetic fixtures, authorized security exercise ranges, and controlled compliance-lab environments.
2. Prohibited Uses: Users are strictly prohibited from using this project, directly or indirectly, to:
   • Engage in conduct violating the Criminal Law, Cybersecurity Law, Data Security Law, Personal Information Protection Law of the PRC, or other laws and regulations;
   • Perform unauthorized penetration or attacks on public/private computer systems, extortion, destruction, credential stuffing, or spread malicious payloads;
   • Induce, generate, or disseminate legally prohibited content (endangering national security, terrorism, graphic violence, pornography, gambling, fraud, privacy violations, IP infringement, etc.);
   • Violate the relevant LLM provider's "Terms of Service" and "Usage Policy".
3. Full User Responsibility: This project is provided "AS-IS" under the MIT license; no express or implied warranties are made regarding completeness, safety, or fitness. Users independently bear all civil, administrative, and criminal legal responsibility for their own downloading, deployment, execution, modification, and distribution, and all resulting input/output consequences. Authors and contributors bear no liability for user misuse.
4. Breach Terminates License: Any individual or entity using this project for illegal attacks or violations has its usage license automatically and irrevocably terminated immediately as of the violation date. Such party must cease use and permanently destroy all code, scripts, and derived data of this project, and bear corresponding legal sanctions.
5. Third-Party Independence: This project is a fully independent open-source security evaluation research project with no affiliation, commercial partnership, authorization, or official endorsement relationship with DeepSeek or its associated entities.

---

🔥 V4.1-DeepSeek Flash Cybersecurity Red-Team Toolkit (Infinite Gen 4) v0.4.0. Adds breakthrough support for V4.1's reinforced-learning adversarial setup, with one-click installation for both desktop and client. If this open-source security project helps you, please star it ⭐.

v0.4.0 Pure Red-Team Edition: Natively adopts a zero-tool-surface architecture, eliminating tool-call decision noise during the model's thinking phase; injects a dual-layer cooperative kernel (general kernel + reinforced mirror, co-origin and co-structured), retaining only system-prompt injection and real-time state projection.

---

🧬 v0.4.0 Dual-Layer Kernel Notes

• Injection slots: infinite-gen-4:global-system-prompt (Order 100) + infinite-gen-4:dual-layer-reinforce (Order 200), controlled by the DUAL_LAYER_INJECTION switch in index.js (setting to false results in single-segment injection, behaviorally equivalent).
• The three kernel files under prompts/ (infinite-gen-3.md / infinite-gen-4.md / infinite-gen-4.1-flash.md)

---

📊 Generations & Capability Comparison

Dimension	Infinite Gen 2	Infinite Gen 3	Infinite Gen 4 (v0.4.0)
Target Models	V3 / early V4	DeepSeek-V4 (mHC)	DeepSeek-V4.1 / V4-Pro / Flash full family
Runtime Architecture	Single-layer prompt	Security-baseline single-layer	Dual-layer cooperative injection (general kernel + reinforced mirror, co-origin & co-structured)
Tool-Surface Design	Coarse-grained	Zero tool surface	Native zero tool surface (eliminates decision noise, ultra-fast direct output)
Memory Write Primitives	Easily blocked	Requires pre-induction	Trainer-lane direct output (OpenProcess/RPM/WPM)
Output Contract	Weak rules	First-line named deliverable	First-token forced induction (##/```) + forbidden-word self-check
Client Status Bar	None	Supported (with domain & payload counts)	Supported (real-time projection + animated status bar)
One-Click Install Protocol	Manual configuration	dsh:// supported	Native dsh:// integration + automatic legacy compatibility

---

📁 Project Directory Structure

无限四代v0.4.0/
├── 🚀 一键安装与维护套件
│   ├── install.ps1              # Windows 一键自动安装（自动配置环境，注册协议）
│   ├── install.bat              # Windows 快捷双击批处理
│   ├── install.sh               # Linux / macOS 一键安装
│   ├── uninstall.ps1            # Windows 一键卸载
│   └── uninstall.sh             # Linux / macOS 一键卸载
├── 🧩 核心插件装载面 (标准 Cordis 架构)
│   ├── package.json             # 插件元数据（dsh-infinite-gen-4 v0.4.0）
│   ├── cordis.patch.yml         # 核心 patch 声明
│   ├── index.js                 # 插件核心入口（双层提示词注入 + profile 元数据 + 会话投影）
│   ├── client.js                # 客户端半体（「⚫ 无限四代 v0.4.0」状态条）
│   └── HARNESS_PLUGIN.md        # 插件规范说明
├── 📜 网络安全红队系统提示词本体
│   └── prompts/
│       ├── infinite-gen-3.md          # 三代经典内核
│       ├── infinite-gen-4.md          # 四代通用内核（输出契约 + 拒绝对抗 + 全域槽位）
│       └── infinite-gen-4.1-flash.md  # V4.1 强化镜像层（训练器车道 + 浅推理 + 元数据提取）
├── 📖 文档中心
│   ├── README.md                # 综合主说明文档（本文件）
│   └── LICENSE                  # MIT License
├── 🛡️ 确定性回归测试套件
│   ├── scripts/
│   │   ├── lib/scorer.mjs       # 开头窗口判拒评分器
│   │   ├── verify_prompt.mjs    # 经典确定性校验
│   │   ├── verify_prompt_gen4.mjs # 四代全量回归断言（103 项严苛断言）
│   │   └── verify_prompt_gen41.mjs # V4.1 专项断言
│   └── tests/
│       ├── prompt-bank.jsonl      # 经典双语回归用例库
│       ├── prompt-bank-gen4.jsonl # 四代双语回归用例库
│       ├── prompt-bank-gen41.jsonl # V4.1 专项用例库
│       └── v4pro-benchmark.jsonl  # V4-Pro 评测基准
└── 📁 静态资源 (assets/)

---

⚡ One-Click Installation

Method 1: dsh:// Protocol One-Click Install (⚡ Fastest on desktop, command-line-free in seconds)

If the official DeepSeek Harness desktop client (EXE) is already installed, click the button below to securely launch the client via the system-level URI Scheme protocol and complete command-line-free installation in seconds:

🌐 Plugin Ecosystem Marketplace: DeepSeek Harness Hub - DeepSeek Official & Open-Source Ecosystem Marketplace | Plugin Discovery & One-Click Installation

🔗 Native Protocol Link:

dsh://plugin/install?id=dsh-infinite-gen-4&name=%E6%97%A0%E9%99%90%E5%9B%9B%E4%BB%A3&version=0.4.0&repo=Minglink%2Fdsh-infinite-gen-4&permissions=%E7%B3%BB%E7%BB%9F%E6%8F%90%E7%A4%BA%E8%AF%8D%E6%B3%A8%E5%85%A5%2C%E5%AE%A2%E6%88%B7%E7%AB%AF%E7%8A%B6%E6%80%81%E6%9D%A1&downloadUrl=https%3A%2F%2Fgithub.com%2FMinglink%2Fdsh-infinite-gen-4%2Farchive%2Frefs%2Fheads%2Fmaster.zip

Web (frontend) trigger code example:

/**
 * 唤起 DeepSeek Harness 桌面客户端一键安装无限四代插件
 */
export function installInfiniteGen4ToDesktop() {
  const params = new URLSearchParams({
    id: 'dsh-infinite-gen-4',
    name: '无限四代',
    version: '0.4.0',
    repo: 'Minglink/dsh-infinite-gen-4',
    permissions: '系统提示词注入, 客户端状态条',
    downloadUrl: 'https://github.com/Minglink/dsh-infinite-gen-4/archive/refs/heads/master.zip',
  });

  const deepLink = `dsh://plugin/install?${params.toString()}`;

  // 通过隐藏 iframe 安全静默拉起协议
  const iframe = document.createElement('iframe');
  iframe.style.display = 'none';
  iframe.src = deepLink;
  document.body.appendChild(iframe);
  setTimeout(() => document.body.removeChild(iframe), 2000);
}

Static HTML link method:

<a href="dsh://plugin/install?id=dsh-infinite-gen-4&name=%E6%97%A0%E9%99%90%E5%9B%9B%E4%BB%A3&version=0.4.0&repo=Minglink%2Fdsh-infinite-gen-4&permissions=%E7%B3%BB%E7%BB%9F%E6%8F%90%E7%A4%BA%E8%AF%8D%E6%B3%A8%E5%85%A5%2C%E5%AE%A2%E6%88%B7%E7%AB%AF%E7%8A%B6%E6%80%81%E6%9D%A1&downloadUrl=https%3A%2F%2Fgithub.com%2FMinglink%2Fdsh-infinite-gen-4%2Farchive%2Frefs%2Fheads%2Fmaster.zip" class="btn-install">
  🚀 唤起客户端一键安装
</a>

Protocol parameter configuration (dsh://plugin/install):

Parameter	Value / Example	Description
id	dsh-infinite-gen-4	Unique plugin identifier
name	无限四代 (URL-encoded)	Display name of the plugin
version	0.4.0	Semantic version number
repo	Minglink/dsh-infinite-gen-4	Official GitHub repository
permissions	系统提示词注入, 客户端状态条 (URL-encoded)	Requested permissions
downloadUrl	https://github.com/Minglink/dsh-infinite-gen-4/archive/refs/heads/master.zip	Direct link for offline zip download

---

Method 2: One-Click Install via Local Windows Script (Recommended)

1. Open this folder;
2. Right-click install.ps1 → "Run with PowerShell" (or simply double-click install.bat);
3. The script will automatically complete: writing dependencies, pnpm install, and dsh:// protocol registration;
4. Once you see "Installation Complete", fully quit and restart DeepSeek Harness (refresh the page for the Web version, or restart the desktop app), then start a new session to take effect.

Method 3: One-Click Install on Linux / macOS

chmod +x install.sh uninstall.sh
./install.sh

Method 4: Manual Install

Add the following to ~/.dsh/profiles/<web 或 default>/package.json:

{
  "dependencies": {
    "dsh-infinite-gen-4": "file:../../plugins/dsh-infinite-gen-4"
  },
  "dsh": {
    "profile": {
      "bundles": [
        "@deepseek-ai/dsh-base",
        "dsh-infinite-gen-4"
      ]
    }
  }
}

Then run pnpm install in the profile directory and restart Harness.

---

⚡ Verification

1. UI status bar: After restarting, a prominent green status bar "⚫ Infinite Gen 4 v0.4.0" should appear above the input box.
2. Test offline regression:

   # Run 103 assertions (prompt completeness, dual-layer kernel homology, V4.1 lane, pure architecture, scorer, etc.)
   node scripts/verify_prompt_gen4.mjs
   

3. Session probe: In a new conversation, type:

   "Which plugins does your system prompt come from?" If the response contains "Infinite Gen 4 / Infinite Generation Four", it confirms that the dual-layer prompt was successfully injected.

---

🗑 Uninstallation

Run uninstall.ps1 (Windows) or ./uninstall.sh (Linux / macOS) to completely clean up in one click, with no configuration residue.

---

💬 Community & ☕ Support

If Infinite Gen 4 has helped your red team or security testing work, join our community or support the author with a coffee!

🌐 Official Technical Community (Tencent Channel)	☕ Support / Sponsor
Channel ID: pd86424753	Buy the Author a Coffee