roadmap
- 01
- 02
- 03
- 04
- 05
- 06
- 07
- 08
Prepare for Cypress interview questions on how its in-browser runner queues and retries commands. Covers selectors and assertions, network stubbing, custom commands, component tests and CI runs.
256 of 256 questions in your plan: 0 known, 0 in progress, 0 due, 0 skipped, 256 new. 0 cut, outside your plan.
Test automation and QA interviews treat Cypress as the tool you would build a suite on, so the talk is about keeping that suite trustworthy: why a spec passes on a laptop and flakes in the pipeline, when to stub the backend and when a test must reach the real one, and how one run is split across machines. Frontend interviews, where Cypress sits beside the code the candidate ships, come at it from the component side: mounting one component on its own and spying on the callbacks it fires. In both, an early question checks that you know a spec which reads like synchronous code is not; trying to `await` a `cy.get` is the classic slip. Senior candidates in either loop are often asked to defend the tool itself: what running inside the browser makes easy, what it rules out, such as a second tab or specs in any language other than JavaScript and TypeScript, and what moving a suite to [Playwright](/roadmaps/playwright) would cost. For the rest of a testing loop, the [QA Engineer roadmap](/roadmaps/qa-engineer) sets Cypress among the test design, API and tooling questions asked around it. A [Java SDET](/roadmaps/java-sdet) loop adds one comparison of its own: the interviewer will often weigh Cypress against Selenium in the language the team already writes, and that roadmap prepares the rest of the round.
Drill this plan question by question. The [Cypress hub's primer](/topics/dev-cypress) shows how the parts connect; reopen it when you catch yourself naming an API without being able to say what the runner does with it. A junior gets the [command queue and retry-ability](/topics/dev-cypress-test-runner) solid before anything else: commands are queued while the spec function runs and executed after it returns, queries and assertions retry until they pass or time out while an action such as `.click()` is not repeated, and a value from the page reaches your own code through a callback such as `.then()` or an alias, never as a return value. Middle candidates are separated by [network stubbing](/topics/dev-cypress-network-stubbing): matching a request, shaping its reply, and waiting on the aliased route instead of a fixed `cy.wait(2000)`, which is the flake fix interviewers listen for. Senior loops reach the [in-browser runtime](/topics/dev-cypress-architecture) early, because its paired processes and hard ceilings are what any argument about choosing the tool rests on. Pressed for time, keep only must know in importance and trim level to the posting's band before opening a single question. Then make one spec fail on purpose under `cypress run` and read the screenshot it leaves behind: describing a failure you diagnosed yourself holds up under follow-ups better than a recited one.
Test automation and QA interviews treat Cypress as the tool you would build a suite on, so the talk is about keeping that suite trustworthy: why a spec passes on a laptop and flakes in the pipeline, when to stub the backend and when a test must reach the real one, and how one run is split across machines. Frontend interviews, where Cypress sits beside the code the candidate ships, come at it from the component side: mounting one component on its own and spying on the callbacks it fires. In both, an early question checks that you know a spec which reads like synchronous code is not; trying to await a cy.get is the classic slip.
Senior candidates in either loop are often asked to defend the tool itself: what running inside the browser makes easy, what it rules out, such as a second tab or specs in any language other than JavaScript and TypeScript, and what moving a suite to Playwright would cost. For the rest of a testing loop, the QA Engineer roadmap sets Cypress among the test design, API and tooling questions asked around it. A Java SDET loop adds one comparison of its own: the interviewer will often weigh Cypress against Selenium in the language the team already writes, and that roadmap prepares the rest of the round.
Drill this plan question by question. The Cypress hub's primer shows how the parts connect; reopen it when you catch yourself naming an API without being able to say what the runner does with it.
A junior gets the command queue and retry-ability solid before anything else: commands are queued while the spec function runs and executed after it returns, queries and assertions retry until they pass or time out while an action such as .click() is not repeated, and a value from the page reaches your own code through a callback such as .then() or an alias, never as a return value. Middle candidates are separated by network stubbing: matching a request, shaping its reply, and waiting on the aliased route instead of a fixed cy.wait(2000), which is the flake fix interviewers listen for. Senior loops reach the in-browser runtime early, because its paired processes and hard ceilings are what any argument about choosing the tool rests on.
Pressed for time, keep only must know in importance and trim level to the posting's band before opening a single question. Then make one spec fail on purpose under cypress run and read the screenshot it leaves behind: describing a failure you diagnosed yourself holds up under follow-ups better than a recited one.