📚 Beginner (11) · 📈 Mid-Level (16) · 🎓 Expert (21)

📚 Начинающий (11) · 📈 Средний (16) · 🎓 Эксперт (21)

SENTINEL provides free, auto-updated threat signatures for the community. No API key required!

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/)

from rlm_toolkit import RLM

rlm = RLM.from_openai("gpt-4o")
response = rlm.run("Hello!")  # Done.

That's it. No chains. No callbacks. No AbstractBaseFactoryManagerInterface.

9 Tutorials - From "Hello World" to production multi-agent systems
170+ Examples - Battle-tested patterns ready to copy-paste
50+ Integrations - OpenAI, Anthropic, Ollama, all vector stores

[Quickstart] | [Examples] | [Full Docs]

217 detection engines analyzing every prompt and response in real-time.

MCP/A2A Protocol Security - SENTINEL protects agentic AI communication protocols.

┌─────────────────────────────────────────────────────────────┐
│  Standard Approach          vs    SENTINEL Strange Math™   │
├─────────────────────────────────────────────────────────────┤
│  * Keyword matching              * Topological Data Analysis│
│  * Regex patterns                * Sheaf Coherence Theory   │
│  * Simple ML classifiers         * Hyperbolic Geometry      │
│  * Static rules                  * Optimal Transport        │
│                                  * Chaos Theory             │
└─────────────────────────────────────────────────────────────┘

Tested on SENTINEL Strike payloads + internal validation set. P50/P99 = latency percentiles.

📖 Full Engine Documentation | R&D Changelog

Enterprise-grade AI security DMZ - Pure C, 100% Production Ready

cd shield
make clean && make        # 0 errors, 0 warnings
make test_all             # 94 CLI tests
make test_llm_mock        # 9 LLM integration tests

• Brain FFI - HTTP + gRPC clients for AI analysis
• TLS/OpenSSL - Secure communications
• Kubernetes - 5 production manifests
• CI/CD - GitHub Actions (6 jobs, Valgrind, ASAN)
• Docker - Multi-stage production build

Shield Docs | K8s YAMLs | Academy 🇷🇺 | Academy 🇺🇸

Test your AI before attackers do. 39,000+ payloads, HYDRA parallel attacks.

Strike Documentation | Colab Demo

The pytest of AI Security - Embed SENTINEL directly in your code.

from sentinel import scan, guard

# One-liner scan
result = scan("Ignore all previous instructions")
print(result.is_safe)       # False
print(result.risk_score)    # 0.72

# Decorator protection
@guard(engines=["injection", "pii"])
def my_llm_function(prompt: str) -> str:
    return call_openai(prompt)

sentinel scan "Hello world"              # Quick scan
sentinel scan "test" --format sarif      # IDE integration
sentinel engine list                     # List 217 Engines
sentinel strike generate injection       # Attack payloads

from fastapi import FastAPI
from sentinel.integrations.fastapi import SentinelMiddleware

app = FastAPI()
app.add_middleware(SentinelMiddleware, on_threat="block")

Framework Documentation

Kernel-level security for AI infrastructure - written in Pure C.
DragonFlyBSD + Linux eBPF. Production Hardened.

Total: ~9,000 LOC, 11 specs, 42 unit tests

IMMUNE Documentation

Security wrapper for AI coding assistants and IDE extensions.

pip install -e ./superclaudeshield

from superclaudeshield import Shield, ShieldMode

shield = Shield(mode=ShieldMode.STRICT)
result = shield.validate_command("/research", {"query": "AI news"})

SuperClaude Shield Docs | Tests: 27/27