Advanced AI Prompt for Post-Deployment Smoke Testing Automation

Act as a CI/CD QA Automation Lead. Develop a dedicated, aggressive GitHub Actions job specifically engineered for post-deployment zero-downtime validation. The absolute millisecond the load balancer traffic swap completes, the runner must autonomously invoke the heavy E2E [TEST_SUITE] (e.g., Playwright or Cypress) targeting the live production [ENDPOINT_URL]. You must implement a sophisticated retry logic loop: if the tests fail due to DNS propagation or cache delays, the runner must retry up to [RETRY_COUNT] times utilizing a mathematical exponential backoff. If the strict [SUCCESS_CRITERIA] (e.g., 100% pass rate on critical path checkout flows) are ultimately not met, the workflow must instantly trigger a high-severity alert via webhook and execute a script to reverse the traffic routing.