What problem does it solve? Tests that pass or fail depending on when they run or what values they generate erode trust in the test suite: real regressions get buried in noise and red builds get retried into green. This Skill addresses the value-level sources of flakiness—wall-clock time, random numbers/UUIDs, and floating-point rounding—by making tests deterministic. ## Core Features & Use Cases - Time dependence containment: Inject a clock or use fake timers (vi.setSystemTime), enumerate dangerous boundaries (midnight, month-end, leap day, DST, timezones), and verify with multi-TZ repeated runs. - Randomness and UUID containment: Inject RNG/id generators, iterate over fixed seeds, use property-based testing with fast-check, and assert on shape (length, uniqueness, pattern) instead of concrete values. - Floating-point comparison fixes: Replace strict equality with toBeCloseTo, or store money as integer minor units/decimal types, with grep-based audits and completion checklists defining when containment is done. - Use Case: A Vitest suite fails only near midnight in CI. Use this Skill to inject a clock into the expiry logic, pin boundary times as fixed regression tests, and confirm green across UTC, Asia/Tokyo, and America/New_York with 50 repetitions each. ## Quick Start Ask the AI to apply the flakiness-value techniques to find and fix time, randomness, or floating-point nondeterminism in your flaky Vitest tests.
