Flaky Test Detector

Detect, analyze, and eliminate flaky tests across any test runner. An AI agent skill that identifies non-deterministic test failures, categorizes root causes, and applies targeted fixes.

Compatible with opencode, Claude Code, Cursor, Windsurf, and GitHub Copilot.

What it does

• Detects your test runner (Vitest, Jest, Playwright, Cypress, Mocha, AVA)
• Harvests flaky candidates by running tests multiple times
• Analyzes root causes across 5 categories:
  • Async timing issues
  • Shared mutable state
  • Test isolation gaps
  • Environment sensitivity
  • Race conditions
• Applies category-specific fixes with proven strategies
• Verifies fixes by re-running the detection loop
• Generates a flaky test report

Install

# rolecraft — install from GitHub (recommended)
npx rolecraft install sametcelikbicak/flaky-test-detector

# or via agentskill.sh
/learn @sametcelikbicak/flaky-test-detector

# or copy manually
cp -r flaky-test-detector .opencode/skills/

Supported runners

• Vitest
• Jest
• Playwright
• Cypress
• Mocha
• AVA
• react-scripts (CRA)

License

MIT