Terraform provider for Genesys Cloud — managing queues and skills as code

NICE CXone API & Integrations
1

If you are managing queues and skills as Terraform code, you must ensure the Terraform state file itself is encrypted at rest.

The state file contains sensitive org-level configuration data including queue names, skill mappings, and OAuth client IDs. Under our banking compliance framework, storing this state in an unencrypted S3 bucket would be a Dodd-Frank violation. Use S3 server-side encryption with KMS and restrict bucket access via IAM policies.

2

We use Terraform to deploy Predictive Routing configurations alongside our queue definitions.

The ML model binding is a separate Terraform resource, so you can A/B test by deploying two versions of the queue - one with predictive routing and one without - and use feature flags to toggle traffic between them. The Terraform plan diff clearly shows which routing algorithm is active per environment.

3

Among the CCaaS vendors I evaluate, Genesys Cloud has the most mature Infrastructure-as-Code story.

NICE CXone has no official Terraform provider. Five9 has a REST API but no IaC tooling. Talkdesk offers a beta Terraform provider but it covers only a fraction of their resource types. GC’s Terraform provider covers queues, skills, routing, users, flows, and even Data Tables - it is genuinely production-ready.

4

We integrated the Terraform deployment pipeline with ServiceNow Change Management.

Before terraform apply runs in our CI/CD pipeline, it automatically creates a Change Request in ServiceNow via webhook. The Change Request captures the Terraform plan diff as the technical description. Only after the Change Advisory Board approves the SNOW ticket does the pipeline execute the apply step.

5 
# Complete queue + skill definition
resource "genesyscloud_routing_skill" "billing" {
  name = "Billing_Support"
}

resource "genesyscloud_routing_queue" "billing_q" {
  name                = "Billing Queue"
  description         = "Managed by Terraform"
  acw_wrapup_prompt   = "MANDATORY_TIMEOUT"
  acw_timeout_ms      = 30000
  skill_evaluation_method = "BEST"
  
  members {
    user_id  = data.genesyscloud_user.agent1.id
    ring_num = 1
  }
}

Version control everything. Review queue changes via PR. Never click the admin UI.

6

After every Terraform deployment, I run an analytics validation query to ensure the new queue is receiving traffic.

If the terraform apply succeeds but the queue was accidentally deployed with zero members, the analytics API will show 0 interactions offered. I built a post-deployment smoke test that queries POST /api/v2/analytics/queues/observations/query and alerts if any managed queue shows zero activity within 30 minutes of deployment.

7

As an AppFoundry partner, we actually distribute our GC integrations as Terraform modules.

Customers add our module to their Terraform configuration, run terraform apply, and our integration is fully provisioned - OAuth clients, Data Actions, notification channels, and webhook endpoints all deployed in one step. It reduces our average onboarding time from 3 days to 15 minutes.