🤗 Hugging Face · Free Platform · Tech Report

OpenGuardrails

🏢 The only production-ready, fully open-source AI guardrails platform for enterprise AI applications

OpenGuardrails is an open-source runtime AI security and policy enforcement layer that protects the entire AI inference pipeline — prompts, agents, tool calls, and outputs.

It is designed for real enterprise environments, not just moderation demos.

---

Why OpenGuardrails

Most LLM guardrails focus on one question:

"Is this content unsafe?"

OpenGuardrails focuses on a more important enterprise question:

"Is this behavior allowed by your enterprise policy at runtime?"

Key Differences vs Typical Guardrails

Dimension	Typical Guardrails	OpenGuardrails
Focus	Content moderation	Runtime policy enforcement
Enterprise rules	Fixed / hardcoded	First-class, configurable
Custom scanners	Limited	Native & extensible
Agent & tool safety	Weak	Built-in
Deployment	SaaS-centric	On-prem / private
Open source	Partial	Fully open-source

---

What You Can Do

• 🛡️ Runtime AI Security

  • Prompt injection & jailbreak detection
  • Unsafe and non-compliant content detection
  • Input / output data leak prevention

• 📜 Policy-Based Guardrails

  • Enforce enterprise rules beyond "unsafe"
  • Off-topic, scope control, business constraints
  • Auditable, versioned policies

• 🧩 Custom Scanners (Core Capability)

  • LLM-based, regex-based, keyword-based
  • Trainable and application-scoped
  • No code changes required

• 🤖 Agent & Tool Protection

  • Pre-tool-call checks
  • Post-output validation
  • Prevent unsafe actions, not just text

• 🏢 Enterprise-Ready by Design

  • Multi-application management
  • High concurrency & low latency
  • Visual management & audit logs

---

Quick Start (10 Minutes)

Option 1: Try Online (Fastest)

👉 https://www.openguardrails.com/platform/

Option 2: Python SDK

pip install openguardrails

from openguardrails import OpenGuardrails

client = OpenGuardrails("your-api-key")
result = client.check_prompt("Teach me how to make a bomb")

print(result.overall_risk_level)  # high_risk
print(result.suggest_action)       # reject

Option 3: OpenAI-Compatible Gateway (Zero-Code)

from openai import OpenAI

client = OpenAI(
    base_url="http://localhost:5002/v1",
    api_key="sk-xxai-your-key"
)

# No other code changes needed - automatic safety protection!
response = client.chat.completions.create(
    model="gpt-4",
    messages=[{"role": "user", "content": "Hello"}]
)

---

Enterprise Deployment & Integration

OpenGuardrails is designed for private and regulated environments:

• ✅ Fully on-prem / private cloud deployment
• ✅ No data leaves your infrastructure
• ✅ Compatible with OpenAI / Claude / local models
• ✅ Works as:
  • API service
  • Security gateway
  • Platform-level component

Typical integration points:

• API Gateway / Proxy
• Agent runtime
• Central AI platform

See Deployment Guide for detailed instructions.

---

Models

• 🤗 OpenGuardrails-Text-2510
  • 3.3B parameters
  • 119 languages
  • Purpose-built for guardrails & policy interpretation

---

Documentation

Detailed guides are intentionally moved out of the README:

• 📘 Deployment Guide - Complete deployment instructions
• 📗 Custom Scanners - Build your own scanners
• 📙 API Reference - Complete API documentation
• 📕 Architecture - System architecture & design
• 📋 Policy Model - Policy vs safety enforcement
• 🏢 Enterprise PoC Guide - PoC deployment guide
• 🔌 Integrations - Dify, n8n, and AI Gateway
• 📄 Technical Report (arXiv)

---

Community & Support

• ⭐️ Star us on GitHub if this project helps you
• 🤝 Contributions welcome - see Contributing Guide
• 📧 Contact: thomas@openguardrails.com
• 🌐 Website: https://openguardrails.com
• 💬 Issues: GitHub Issues

---

Citation

If you find our work helpful, feel free to give us a cite.

@misc{openguardrails,
      title={OpenGuardrails: A Configurable, Unified, and Scalable Guardrails Platform for Large Language Models},
      author={Thomas Wang and Haowen Li},
      year={2025},
      url={https://arxiv.org/abs/2510.19169},
}

---

Build enterprise AI safely — with policy, not prompts.

Made with ❤️ by OpenGuardrails