New Integration: PolicyAI Content Moderation

[vitchor]

Hi NeMo Guardrails team! 👋

First off, thank you for building such an excellent toolkit — we've started using NeMo Guardrails at Musubi Labs and love how easy it makes adding programmable guardrails.

We've put together a contribution to add PolicyAI as a new third-party integration for content moderation, following the same pattern as the ActiveFence integration.

PR: #1576

What is PolicyAI?

PolicyAI is a policy-based content moderation API that lets teams:

• Define custom policies in natural language and rapidly iterate through editing, evaluation, and deployment
• Organize policies with tags to easily switch between groups/versions of policies
• Cluster & explore flagged content to detect novel emerging attacks
• Get structured decisions with assessment, category, severity, and reasoning

What's included in the PR?

• New integration module at nemoguardrails/library/policyai/
• Colang 1.x and 2.x flows for input/output moderation
• Documentation in the community guides and guardrail catalog
• Comprehensive test coverage

We'd love any feedback on the implementation. Happy to make adjustments based on your guidance!

Thanks for your time 🙏