Raegis Scientific Audit Station 🔬 HQ Edition

The Deep-Analysis Engines for the Raegis Protocol

While the Raegis-JS SDK handles lightweight UI and monitoring, this station provides the Heavy Scientific Inference required for deep-learning auditing of models like Google Gemma and Llama.

📐 Scientific Features

• Whitebox Inspection: Real-time token entropy extracted from local model weights (PyTorch/Transformers).
• Deep Semantic Rivelry: Parallel Sentence-Transformer scoring for RAG fidelity.
• Neural Guardian HQ: GPU-accelerated Autoencoders for anomaly fingerprinting.

🚀 Setup & Launch

1. pip install -r requirements.txt
2. Configure your .env (Set RAEGIS_API_KEY)
3. uvicorn main:app --reload

🐳 Docker Mode

docker build -t raegis-hq-station .
docker run -p 8000:8000 raegis-hq-station

Connecting to the JS SDK:

const audit = await raegis.fullAudit({
  prompt: "...",
  bridgeUrl: "http://localhost:8000",
  apiKey: "YOUR_HQ_API_KEY"
});

Built for the Gemma ecosystem and Jason Mayes' Web-AI vision. By Lucas Frischeisen. LinkedIn