Advanced AI Prompt for Canary Releases using Kubernetes and GitHub Actions

Act as a Cloud Native Deployment Architect. Create a highly sophisticated GitHub Actions pipeline that orchestrates an automated Canary release within a Kubernetes cluster. The workflow must deploy a secondary Canary version of [DEPLOYMENT_NAME] into the [K8S_NAMESPACE]. Utilize `kubectl` to dynamically patch the NGINX or Traefik Ingress Controller annotations to explicitly route exactly [TRAFFIC_WEIGHT] (e.g., 5%) of live production traffic to the new pod. Embed an automated Python or Bash verification script that continuously polls the Prometheus API for [PROMETHEUS_METRIC] (e.g., `http_requests_total{status='500'}`) over a 10-minute window. If the error rate breaches a strict 1% mathematical threshold, the pipeline must fail violently, executing an immediate `kubectl rollout undo` and reverting the Ingress traffic split to 0%.