YONGJAE LEE(이용재)

Use when Cypress tests have actually failed and you need to diagnose runtime failures — from mochawesome or JUnit report files, local or CI. Triggers on "debug cypress tests", "why did cypress tests fail", "cypress CI failure", "flaky cypress test failures", "cypress timed out retrying", "cypress tests pass locally but fail in CI", "analyze cypress/reports". Classifies runtime failures into root causes (not static code analysis) and suggests concrete fixes.

Use when reviewing, auditing, or improving E2E test specs for Playwright or Cypress — static code analysis of existing test files, not diagnosing runtime failures. Triggers on "review my tests", "audit test quality", "find weak tests", "my tests always pass but miss bugs", "tests pass CI but miss regressions", "improve playwright tests", "improve cypress tests", "check test coverage gaps", "my tests are fragile", "tests break on every UI change", "test suite is hard to maintain", "we have coverage but bugs still slip through", "flaky tests", "test anti-patterns", "check my e2e tests", "tests pass locally but fail in CI". Detects 13 anti-patterns -- name-assertion mismatch, missing Then, error swallowing (.catch in POM via grep; try/catch in specs via LLM; Cypress uncaught:exception suppression), always-passing assertions (one-shot booleans, Locator-as-truthy, toBeAttached, timeout:0, one-shot URL), bypass patterns (conditional assertions + force:true), raw DOM queries, focused test leak (test.only committed), missing assertions (dangling locators + boolean result discarded), hard-coded sleeps (P1), flaky test patterns (positional selectors + serial ordering), YAGNI + zombie specs (unused POM members, single-use Util wrappers, zombie spec files), expect.soft() overuse. Also runs supplementary grep checks for general code quality issues (missing auth setup, inconsistent POM usage, hardcoded credentials, missing await, deprecated page API, networkidle).

Use when Playwright tests have actually failed and you need to diagnose runtime failures — from a playwright-report directory, local or CI. Triggers on "debug playwright tests", "why did playwright tests fail", "playwright CI failure", "flaky playwright test failures", "playwright timeout error", "tests pass locally but fail in CI", "analyze playwright-report", "PR failing in CI". Classifies runtime failures into root causes (not static code analysis) and suggests concrete fixes.

Use when generating new Playwright E2E tests from scratch. Triggers on "generate playwright tests", "write e2e tests for X", "add playwright coverage for X", "create test for X page", "generate tests for the login page". Autonomous mode starts from coverage gap analysis when no target is specified; argument mode targets a specific page or feature directly. Explores the live app via Playwright CLI or agent-browser, designs scenarios with user approval via Plan Mode, auto-detects project structure (POM vs flat spec), runs YAGNI audit and e2e-reviewer after generation, and hands off to playwright-debugger after 3 failed fix attempts.