Hooks: before, beforeEach, after, afterEach

What was run

A real Cypress spec recording its own hook-execution order into a shared array, then asserting on the exact real sequence in an after() hook — directly measured, not assumed from documentation.

describe("hooks execution order", () => {
  const order = [];

  before(() => { order.push("before"); });
  beforeEach(() => { order.push("beforeEach"); });
  afterEach(() => { order.push("afterEach"); });
  after(() => {
    order.push("after");
    expect(order).to.deep.eq([
      "before",
      "beforeEach", "test1", "afterEach",
      "beforeEach", "test2", "afterEach",
      "after",
    ]);
  });

  it("test1", () => { order.push("test1"); });
  it("test2", () => { order.push("test2"); });
});

✓ test1 (17ms)
✓ test2 (2ms)
2 passing (35ms)

Both tests passed, and critically, the after() hook's own real deep-equality assertion against the recorded order also passed silently — confirming the real execution sequence was exactly before once, then beforeEach/test/afterEach repeated for each test, then after once at the very end.

A real, deliberately broken order

// changed the after() hook's expected first entry to "WRONG_ORDER"

1) "after all" hook for "test2":
   AssertionError: expected [ Array(8) ] to deeply equal [ Array(8) ]

Because this error occurred during a `after all` hook we are skipping the remaining
tests in the current suite: `hooks execution order`

Cypress/Mocha genuinely reports this failure as attached to "after all" hook for "test2", even though the hook itself runs once for the whole suite, not per-test — and it explicitly states its own real policy of skipping any remaining tests in the suite once an after all hook throws.

How It Actually Works

before/after run exactly once per describe block, no matter how many tests it contains; beforeEach/afterEach run once per individual test. This lesson's real recorded array is the direct proof: "before" and "after" each appear exactly once, while "beforeEach"/"afterEach" each appear twice — once per real test. Hooks run as genuine Cypress commands themselves (they can contain cy.visit(), cy.get(), and so on, exactly like a test body), which is why this course's earlier login-flow patterns could be moved into a beforeEach to avoid repeating a login sequence in every single test, the same real motivation behind custom commands — hooks and custom commands solve overlapping but distinct problems (hooks for when setup runs, commands for what the setup does).

Common mistakes

• Assuming a failed after() hook only affects reporting, not subsequent tests — this lesson's real error message states plainly that remaining tests in the current suite are skipped once an after all hook throws, which can silently hide real test failures that would have been caught by tests that never ran.
• Putting assertions that depend on test-specific state inside after() instead of afterEach() — after() runs once, after every test in the block has already finished, so it can only sensibly assert on cumulative state (like this lesson's full order array), not a single test's own result.
• Forgetting that beforeEach/afterEach apply to every test in their enclosing describe block, including ones added later — adding a tenth test to this lesson's suite would make it run beforeEach/afterEach a tenth time automatically, without any change to the hook definitions themselves.

Exercise

Nest a second describe() block inside this lesson's existing one, with its own beforeEach() that pushes "inner-beforeEach". Add one test inside the nested block, run the suite, and determine from the real resulting order array whether Cypress runs outer and inner beforeEach hooks for that nested test in outer-then-inner order, inner-then-outer, or only the innermost one — directly observing Mocha's real hook nesting behavior rather than assuming it.