Glossary
QA terms, without the jargon.
Short, practical definitions for the words that show up in QA conversations. No circular dictionary entries; each definition explains why the concept matters.
Last updated October 31, 2026
A
- Autonomous testing
- Autonomous testing describes systems that discover application behavior and choose their own paths without a human authoring each test. Verification is what separates it from random clicking.
- Exploratory testingAI QA agentBug verification
- AI QA agent
- An AI QA agent uses a model plus browser automation to explore an application, reason about expected behavior and report issues. The strongest implementations verify findings before reporting them.
- Autonomous testingExploratory testing
B
- Bug verification
- Verification replays the steps that produced a suspicious observation in a clean environment. Locodera requires 3/3 successful reproductions before a finding is confirmed.
- False positiveReproduction rate
- Browser automation
- Browser automation uses tools such as Playwright or Cypress to drive a real or headless browser. Locodera uses Playwright as its execution layer.
- End-to-end testingHeadless browser
- Bug report
- A useful bug report includes steps to reproduce, expected and actual behavior, environment, severity and evidence. It should not assert a root cause without proof.
- SeverityBug verification
C
- CI/CD
- CI/CD automates building, testing and deploying software. Post-deploy QA hooks run checks against the newly deployed environment.
- Deploy webhookRegression testing
D
- Deploy webhook
- A deploy webhook lets CI/CD trigger work in another system. Locodera's per-project webhook queues a full exploratory scan after a release.
- CI/CDAutonomous testing
E
- Exploratory testing
- Exploratory testing is simultaneous learning, test design and execution. The tester uses the product, observes behavior and decides the next action based on what they see, rather than following a fixed script.
- Test caseRegression testingAutonomous testing
- End-to-end testing
- End-to-end (E2E) testing exercises a flow from the user interface through the backend, verifying the whole path. It is powerful but expensive to author and maintain.
- Test caseBrowser automation
F
- False positive
- A false positive is an issue report that cannot be reproduced or does not represent a defect. Repeated verification and clean-context replays are the main defenses.
- Bug verificationReproduction rate
- Flaky test
- A flaky test produces inconsistent results because of timing, environment or ordering. Flakiness erodes trust; repeated clean-context verification is one way to distinguish it from a real intermittent bug.
- Reproduction rateFalse positive
H
- Headless browser
- A headless browser runs the rendering and JavaScript engine without a UI. It is the standard way to run browser tests in CI.
- Browser automationCI/CD
R
- Reproduction rate
- The reproduction rate is the fraction of attempts in which a bug occurs. Locodera reports confirmed issues as 3/3; intermittent observations below that threshold are discarded.
- Bug verificationFlaky test
- Regression testing
- Regression testing re-runs known checks after a change to catch new breakage. It protects known behavior; exploratory testing looks for behavior the suite does not cover.
- Test caseSmoke testing
- Release risk score
- A release risk score combines confirmed severity, reproducibility, regressions and passed flows into one decision aid. Locodera calls its version the Don't Ship Score.
- SeverityRegression testing
S
- Smoke testing
- Smoke testing is a shallow pass over critical functionality to decide whether a build is worth deeper testing. It is not a substitute for exploratory or regression coverage.
- Regression testingSanity testing
- Sanity testing
- Sanity testing verifies a narrow area after a change, usually without a full regression run. It answers 'did this specific fix work?' rather than 'is the whole product healthy?'.
- Smoke testingRegression testing
- Severity
- Severity describes the impact of a defect on the product or user. It is distinct from priority, which describes how urgently the team intends to fix it.
- Bug reportRelease risk score
T
- Test case
- A test case is a repeatable check with a setup, steps, input and expected result. Test suites are valuable for known behavior but only cover scenarios someone thought to write.
- Regression testingTest oracle
- Test oracle
- A test oracle is the mechanism used to judge a result: an assertion, a specification, an invariant or a reference implementation. Without an oracle, observing behavior is not testing.
- Test caseBug verification
Put the terms to work
The definitions are useful, but the real question is whether your app behaves the way it promises. Run an exploratory scan and see.
Join the waitlistFind the bugs your tests never looked for.
Add your app, run an exploratory scan, and get a release verdict backed by evidence.