Vocabulary for CI/CD Pipelines: Build, Test and Deploy Terms Explained

Master the English vocabulary of CI/CD pipelines — stages, jobs, artifacts, flaky tests, rollbacks and deployment strategies — with example sentences for daily standups.

CI/CD pipelines are where a lot of your daily English happens: “the build’s red,” “it’s stuck in the deploy stage,” “that test is flaky again.” The vocabulary is specific, and using the wrong word makes you sound less fluent than you are. This guide covers the build, test, and deploy vocabulary you need, with example sentences for standups and PRs.


The big two: CI and CD

  • CI — Continuous Integration — automatically building and testing every change. “CI runs on every push.”
  • CD — Continuous Delivery / Deployment — automatically getting changes to staging or production. “CD deploys to staging on merge.”

Note the difference: Continuous Delivery means ready to deploy at the click of a button; Continuous Deployment means deployed automatically with no manual step. People say “CD” for both — clarify if it matters.


Pipeline structure

  • pipeline — the whole automated sequence. “The pipeline takes twelve minutes.”
  • stage — a phase (build, test, deploy). “It’s stuck in the deploy stage.”
  • job — a unit of work within a stage. “The lint job failed.”
  • step — a single command in a job.
  • runner / agent — the machine that executes jobs. “No runners are free.”
  • workflow — the configured pipeline (GitHub Actions term). “I edited the workflow file.”
  • to trigger — to start a pipeline. “The push triggered a build.”
  • trigger / on push / on merge — what starts it. “This workflow triggers on PRs.”

“The pipeline has three stages — build, test, deploy — and the test stage runs four jobs in parallel.”


Build vocabulary

  • build — compiling/packaging the code. “The build’s green.”
  • to compile / to package“It compiles but the package step fails.”
  • artifact — the output of a build (a binary, image, or bundle). “The build artifact is a Docker image.”
  • cache — stored data to speed up builds. “The dependency cache is stale.”
  • to fail fast — stop at the first error. “Configure it to fail fast.”
  • green / red — passing / failing. “Main is red again.”
  • broken build — a build that fails. “Who broke the build?”
  • to bisect — narrow down which commit broke something. “I bisected it to last night’s merge.”

Test vocabulary

  • test suite — the full set of tests. “The suite takes eight minutes.”
  • unit / integration / end-to-end (E2E) tests — increasing scope. “E2E is the slow part.”
  • flaky test — passes and fails non-deterministically. “That test is flaky — re-run it.”
  • to flake“It flaked again on CI but passes locally.”
  • coverage — how much code the tests exercise. “Coverage dropped below 80%.”
  • to skip / to quarantine — temporarily disable a test. “I quarantined the flaky one.”
  • regression test — guards against a bug coming back.
  • smoke test — a quick check that the basics work. “Smoke tests pass.”
  • to re-run / to retry“Re-run the job; it’s probably a flake.”

“The unit tests pass, but two E2E tests are flaky and keep going red on CI even though they’re green locally.”

The word flaky (and the verb to flake) is one of the most used terms in CI — learn it cold.


Deploy vocabulary

  • to deploy / to ship / to release / to roll out — push to an environment. “We’re shipping at 3pm.”
  • environment — staging, prod, QA. “It’s live in staging.”
  • to promote — move a build from one environment to the next. “Promote it to prod.”
  • rollout — the act of deploying. “The rollout is at 50%.”
  • to roll back / revert — undo a deploy. “Roll back to the last good version.”
  • canary — release to a small subset first. “It’s in canary, looking healthy.”
  • blue-green — two identical environments, switch traffic between them.
  • feature flag / toggle — turn features on/off without deploying. “It’s behind a flag.”
  • smoke test on deploy — post-deploy sanity check.

Deployment strategies in one line each:

  • Canary“Send 5% of traffic to the new version and watch.”
  • Blue-green“Deploy to the idle environment, then flip the switch.”
  • Rolling“Replace instances a few at a time.”

When things go wrong

  • broken pipeline“The pipeline’s broken for everyone.”
  • stuck / hung — not progressing. “The deploy job’s hung.”
  • timeout — exceeded the time limit. “The job timed out.”
  • flaky infra — the runners themselves are unreliable.
  • to unblock — clear what’s stopping the pipeline. “I restarted the runner to unblock it.”
  • to gate — block progress until a condition is met. “Merges are gated on green CI.”

“Main’s red, but it’s not the code — a runner died and the deploy job timed out. I re-ran it and it’s green now.”


Standup and PR phrases

  • “CI’s green; this is ready to merge.”
  • “The build’s red on a flaky test — re-running.”
  • “It’s deployed to staging; I’ll promote to prod after smoke tests.”
  • “Rollout’s paused at 25% — error rate ticked up.”
  • “I rolled it back; we’ll re-deploy once the fix lands.”
  • “This is gated behind a feature flag, so it’s safe to merge.”

Common mistakes

  • “The test is broken” vs “flaky”. Broken = consistently failing (real bug). Flaky = sometimes failing (unreliable test).
  • Confusing deploy and release. Deploy = code is out; release = users can use it (a flag may still hide it).
  • “Rollback” as one word vs “roll back” the verb. The noun is rollback; the verb is to roll back.
  • Using “build” for the whole pipeline. The build is one stage; the pipeline is the whole thing.
  • “Promote” vs “deploy”. Promote moves an existing build forward; deploy runs the deployment.

Key takeaways

  • Know the structure: pipeline → stages → jobs → steps, run by runners.
  • Build terms: artifact, cache, green/red, fail fast, bisect.
  • Test terms: suite, flaky, coverage, quarantine, smoke test.
  • Deploy terms: promote, roll back, canary, blue-green, feature flag.
  • Distinguish flaky vs broken and deploy vs release — they cause the most confusion.

Master this vocabulary and every standup update about your pipeline will be crisp, accurate, and instantly understood.