Glossary

Plain-English definitions of the terms used across BugBrain. Where a concept has its own page, the term links to it.

A#

Add-on — A metered capability that is off by default and enabled per workspace (its feature flag on and its quota above zero). The add-ons are PR automation, load testing, and AI / LLM app testing. See Plans & quotas.

AI agent — One of the autonomous components that explores your app, drives flows, and judges results. See How the AI works.

Application knowledge graph — BugBrain's learned map of your app: its pages (states) and the actions that move between them. It sharpens test selection and self-healing over time. See Application knowledge graph.

C#

Confidence score — A 0-to-1 measure of how sure the agent is about a step or an issue. Low confidence pushes a result toward INCONCLUSIVE rather than a false FAIL. See Issue confidence & dedup.

E#

Exploratory run — An open-ended run where the AI decides what to test, mapping pages and exercising high-value flows without a script. See Test runs.

F#

Feature flag — A platform switch that turns a capability on or off (and controls rollout). Add-ons and integrations are gated, in part, by feature flags.

Flaky test — A test whose result flips between pass and fail without a real change in the app. BugBrain detects flakiness and can quarantine unstable tests. See Self-healing tests.

I#

Integration — A connection to an external tool — an issue tracker where BugBrain files bugs, or a notifier where it sends alerts. See Integrations.

Issue — A problem a run surfaced (a broken interaction, console error, failed request, accessibility problem, and so on), with severity, category, and evidence. See Issues & bug triage.

M#

MCP server — The Model Context Protocol server that connects AI coding tools (Claude Code, Cursor, Codex) to BugBrain, exposing a scope-gated set of tools. See MCP server.

O#

Oracle — The check that decides whether what the agent observed is correct or a defect — the logic behind each PASS, FAIL, or INCONCLUSIVE verdict. See Test-run scoring.

P#

Premium Agent — BugBrain's premium browser-driving engine, powering premium runs and advanced self-healing. Always referred to as the Premium Agent. See Premium Agent.

Project — A target the agent tests: a reachable URL plus its settings, credentials, and knowledge. See Projects.

R#

Run — A single execution against a project (exploratory, a test case, a plan, scheduled, or triggered by a pull request). See Test runs.

S#

Self-healing — Re-resolving a locator that moved so a test keeps passing through UI change, instead of failing on a brittle selector. See Self-healing tests.

Signature — A stable fingerprint of an issue used to deduplicate it across runs, so the same defect isn't reported many times. See Issue confidence & dedup.

T#

Test case — A saved, repeatable check the agent runs the same way each time. See Test cases.

Test plan — A grouped set of test cases run together. See Test plans.

V#

Verdict (PASS / FAIL / INCONCLUSIVE) — The outcome of a check. PASS: confirmed correct. FAIL: definitively not met. INCONCLUSIVE: the agent couldn't confidently confirm or deny (element not found, CAPTCHA, low confidence) — used instead of forcing a false FAIL. See Test-run scoring.

W#

Workspace — Your organization in BugBrain: the account that holds projects, members, billing, and settings. (Also called the organization.)

Related#