Nyraxis Documentation

Nyraxis is an AI governance platform that evaluates every LLM input and output against configurable security policies — PII detection, prompt injection blocking, toxicity filtering, and more.

What you get

• 16 built-in providers — PII, prompt injection, jailbreak, toxicity, secrets, bias, NSFW, hallucination, and more
• Sub-500ms latency — evaluate in-line without slowing your agent
• Python & Node SDKs — two lines of code to protect any LLM call
• Auto-enforce mode — monkey-patches OpenAI/Anthropic SDKs so every call is evaluated automatically
• Dashboard — real-time visibility into violations, traces, and policy health

Quick example

import nyraxis_sdk

nyraxis_sdk.init(api_key="nyx_...", enforce=True)

# Every OpenAI call is now protected — no wrapper code needed
import openai
client = openai.OpenAI()
client.chat.completions.create(
    model="gpt-4o",
    messages=[{"role": "user", "content": "user input here"}],
)

Architecture

User Input → Your Agent → Nyraxis Evaluate → Allow/Block → LLM
                                ↓
                         Dashboard & Alerts

Nyraxis sits between your agent code and the LLM. It evaluates text against your active policies and either allows the request to proceed or blocks it with a violation report.

Next steps

• Quickstart — from zero to first evaluation in 5 minutes
• Concepts — understand policies, modes, and enforcement
• Python SDK — full Python reference
• Node SDK — full Node.js/TypeScript reference
• Providers — explore all 16 built-in providers