Flaky Test Detective is a skill for AI coding agents, developers, QA engineers, CI/CD teams, and test automation teams who need to investigate tests that pass and fail inconsistently. It covers a broad range of test types: unit tests, integration tests, E2E tests, browser tests, API tests, and database tests, across frameworks including Playwright, Cypress, Selenium, Jest, Vitest, and pytest. It also handles CI-only failures, which are often the hardest class of flakiness to reproduce locally.
For any failing or unreliable test, the skill produces root cause hypotheses and failure pattern reports. It performs async timing audits to identify race conditions, selector stability reviews for browser tests, and mock leakage checks where shared state bleeds between tests. It reviews database cleanup procedures, compares CI versus local environment configurations, and produces test isolation plans. From those findings it generates stabilization roadmaps and QA tickets that describe what to fix and how to verify the fix.
The skill is explicitly designed to avoid masking failures. It does not recommend retries, arbitrary sleep calls, skipped tests, or weakened assertions as solutions. Instead it produces paste-ready prompts formatted for AI coding tools including Cursor, Claude Code, Codex CLI, OpenCode, Replit, and ChatGPT Agents, so the identified root cause can be handed directly to a coding agent for remediation.