This systematically hunts down and fixes flaky tests using Tuist's test insights API. It pulls reliability metrics, analyzes failure patterns across runs, and walks through the actual source code to identify root causes like race conditions, shared state pollution, or missing async waits. The workflow is thorough: it can triage all flaky tests in a project, group them by common causes, and distinguish between actual test bugs versus infrastructure failures. What's refreshing is the pragmatic verification guidance, including when it's okay to fix without reproducing locally if the code analysis is solid. Best suited for projects already using Tuist Cloud for test analytics, where you have structured failure data to work from rather than just gut instinct.
npx skills add https://github.com/tuist/agent-skills --skill fix-flaky-tests