#2363·Guardrails

doc:add Conduct to third-party guardrail catalog

Author: sseshachalaCreated Sep 5, 2026Updated Sep 5, 2026
Labelsdocumentationstatus: needs triage

Please also confirm the following

  • I have searched the main issue tracker of NeMo Guardrails repository and believe that this is not a duplicate

Issue Kind

Improving documentation

Existing Link

https://github.com/NVIDIA-NeMo/Guardrails/blob/develop/docs/configure-rails/guardrail-catalog/third-party.mdx

Description

Add conduct-nemo-guard to the third-party guardrail catalog (docs/configure-rails/guardrail-catalog/community/). Conduct is a runtime policy layer for AI agents; the plugin exposes a conduct_guard_check Colang action that lets input rails call an external policy service and branch on allow / warn / block / approval verdicts — the same integration shape as ActiveFence, CrowdStrike AIDR, Prompt Security, and Pangea already in the catalog.

Motivation

Users running NeMo Guardrails alongside a workspace-wide policy layer (RBAC, audit, approval workflows, cross-agent rules) currently have no documented pattern for wiring an external policy verdict into a Colang input rail. Adding Conduct to the catalog fills that gap and follows the pattern already established for other third-party HTTP-based guards.

Proposed changes

  • New page: docs/configure-rails/guardrail-catalog/community/conduct.mdx
  • One nav entry in docs/index.yml under Third-Party APIs
  • No code changes to nemoguardrails/ — the plugin ships as a separate PyPI package (conduct-nemo-guard)

Links

Contribution
I'll open the PR against develop with DCO sign-off once assigned to this issue, per CONTRIBUTING.md.