How do you diagnose and fix flaky Cypress tests?

A flaky test passes and fails without a code change. That usually means the test depends on something it does not control: a request that may or may not finish, data left by another test, an animation, or a selector that matches more than one element. The fix is to make the dependency explicit.

Common root causes

• Fixed waits that are too short on slow machines
• Requests that race with the assertion
• Shared database records created by another test
• Selectors that match multiple elements and depend on order
• Animations or transitions that intercept a click
• Time, locale or random data that differs between runs
• Order dependence hidden by a single passing run

beforeEach(() => {
  cy.intercept("GET", "/api/orders").as("getOrders");
  cy.visit("/orders");
  cy.wait("@getOrders");
});

it("shows the empty state", () => {
  cy.intercept("GET", "/api/orders", []);
  cy.reload();
  cy.get("[data-cy=empty]").should("be.visible");
});

Reproduce the failure before changing anything. Run the spec repeatedly, run it alone and with the full suite, and check whether the failure follows a specific test. Cypress can retry tests in run mode, but retries are a signal, not a cure. A test that only passes on the second attempt is still broken.