AI agents in test automation: building a workflow from skills, MCP, and different LLMs

▤ Summary

The article explains how AI agents can automate QA testing by chaining skills, prompts, and different LLMs into a workflow. It describes a practical setup that uses Claude, Codex, and MCP-connected tools to manage tasks, estimate work, run tests, and create pull requests. The workflow focuses on reducing manual QA effort across test creation, updates, failure investigation, and flaky test handling. It also shows how task history and structured skills help the agent stay consistent across CI/CD and development processes.

▦ Linked Companies