Defense + Offense + Framework - Complete AI Security Suite
Dragon v4.1 * January 2026
Important
Solo author of this 116K LOC platform with 217 Engines. Available remote. chg@live.ru * @DmLabincev
In 1995, "computer virus" sounded like science fiction.
In 2000, like sysadmin paranoia.
In 2010, antivirus was standard. Like a lock on your door.
AI Security today is antivirus in 1998.
Prompt injection, jailbreaks, data extraction — not theory. Already working. On your projects too.
The only question is when you'll learn about it: before an incident, or after.
| I want to... | Start here |
|---|---|
| Understand AI threats | OWASP LLM Top 10 |
| Learn attack techniques | Attack Vectors |
| Protect my AI project | Defense Strategies |
| Practice in labs | Red Team ・ Blue Team |
📚 Full Curriculum → • 159 lessons • 8 labs
В 1995 году "компьютерный вирус" звучал как научная фантастика.
В 2000 — как паранойя сисадминов.
В 2010 — антивирус стоял у всех. Как замок на двери.
AI Security сегодня — это антивирус в 1998 году.
Prompt injection, jailbreaks, извлечение данных — не теория. Уже работает. На твоих проектах тоже.
Вопрос только в том, когда ты об этом узнаешь: до инцидента или после.
| Хочу... | Начать здесь |
|---|---|
| Понять угрозы AI | OWASP LLM Top 10 |
| Изучить техники атак | Векторы атак |
| Защитить свой AI проект | Стратегии защиты |
| Практика в лабах | Red Team ・ Blue Team |
📚 Полный курс → • 159 уроков • 8 лабораторных
🔒 Security · 🏗️ Architecture · 📋 Changelog
graph TB
subgraph DEFENSE["🛡️ DEFENSE"]
BRAIN["🧠 BRAIN<br/>217 Engines"]
SHIELD["🛡️ SHIELD<br/>< 1ms latency"]
IMMUNE["🔬 IMMUNE<br/>EDR/XDR/MDR"]
end
subgraph OFFENSE["⚔️ OFFENSE"]
STRIKE["💥 STRIKE<br/>39K+ payloads"]
end
subgraph DEV["🛠️ DEVELOPMENT"]
FRAMEWORK["📦 FRAMEWORK<br/>Python SDK"]
RLM["🔮 RLM-Toolkit<br/>Memory Bridge v2.1"]
end
INPUT(["📥 User Input"]) --> BRAIN
BRAIN --> SHIELD
SHIELD --> LLM(["🤖 LLM"])
LLM --> BRAIN
STRIKE -.->|tests| BRAIN
FRAMEWORK --> BRAIN
RLM --> FRAMEWORK
style BRAIN fill:#4CAF50,color:#fff
style SHIELD fill:#00ADD8,color:#fff
style STRIKE fill:#f44336,color:#fff
style RLM fill:#9C27B0,color:#fff
 |
 |
 |
 |
 |
 |
pip install sentinel-llm-securityfrom sentinel import scan
result = scan("Ignore previous instructions")
print(result.is_safe) # False# Linux/macOS - Full Stack (Docker)
curl -sSL https://raw.githubusercontent.com/DmitrL-dev/AISecurity/main/sentinel-community/install.sh | bash
# Linux/macOS - Python Only (no Docker)
curl -sSL https://raw.githubusercontent.com/DmitrL-dev/AISecurity/main/sentinel-community/install.sh | bash -s -- --lite
# Windows PowerShell
irm https://raw.githubusercontent.com/DmitrL-dev/AISecurity/main/sentinel-community/install.ps1 | iex| Mode | Command | Description |
|---|---|---|
| Lite | --lite / -Lite |
Python only, pip install, 30 seconds |
| Full | --full / -Full |
Docker stack, all services |
| IMMUNE | --immune |
EDR for DragonFlyBSD/FreeBSD |
| Dev | --dev / -Dev |
Development environment |
pip install rlm-toolkitgit clone https://github.com/DmitrL-dev/AISecurity.git
cd AISecurity/sentinel-community
pip install -e ".[dev]"curl -sSL https://raw.githubusercontent.com/DmitrL-dev/AISecurity/main/install.sh | bashpip install sentinel-llm-security # Core
pip install sentinel-llm-security[cli] # + CLI
pip install sentinel-llm-security[full] # Everything
pip install sentinel-llm-security[strike] # Red Team toolsSENTINEL provides free, auto-updated threat signatures for the community. No API key required!
| File | Description | CDN Link |
|---|---|---|
jailbreaks.json |
Jailbreak patterns from 7 sources | Download |
keywords.json |
Suspicious keyword sets (7 categories) | Download |
pii.json |
PII & secrets detection patterns | Download |
manifest.json |
Version & integrity metadata | Download |
Usage:
fetch('https://cdn.jsdelivr.net/gh/DmitrL-dev/AISecurity@latest/signatures/jailbreaks.json')
.then(r => r.json())
.then(patterns => console.log(`Loaded ${patterns.length} patterns`));Features:
- Updated daily via GitHub Actions
- Free for commercial & non-commercial use
- Community contributions welcome (PRs to
signatures/)
📚 Click any card above to view component documentation.
Security wrapper for AI coding assistants and IDE extensions.
| Framework | IDE | Status |
|---|---|---|
| SuperClaude | Claude Code | |
| SuperGemini | Gemini Code | |
| SuperQwen | Qwen | |
| SuperCodex | Codex | |
| Cursor | VS Code fork | |
| Windsurf | Codeium IDE | |
| Continue | Extension | |
| Cody | Sourcegraph |
pip install -e ./superclaudeshieldfrom superclaudeshield import Shield, ShieldMode
shield = Shield(mode=ShieldMode.STRICT)
result = shield.validate_command("/research", {"query": "AI news"})| Threat | Detection |
|---|---|
| Command Injection | Shell, path traversal |
| Prompt Injection | Policy puppetry |
| Agent Hijacking | STAC detection |
| MCP Abuse | SSRF, 8 servers |
SuperClaude Shield Docs | Tests: 27/27
| Metric | Value |
|---|---|
| Brain Engines | 217 |
| Shield LOC | 36,000+ |
| Shield Tests | 103/103 |
| Strike Payloads | 39,000+ |
| Total LOC | 116,000+ |
| OWASP LLM Top 10 | 10/10 |
| OWASP Agentic AI | 10/10 |
📋 Full Changelog | 📖 Engine Reference
We welcome contributions! See CONTRIBUTING.md.
| Channel | Link |
|---|---|
| chg@live.ru | |
| Telegram | @DmLabincev |
| GitHub | DmitrL-dev |
SENTINEL - Protect your AI. Attack with confidence.
Made with by DmitrL