Skip to main content

Glossary

This page is the single source of truth for what we call things in msg.ZenTestAI. Several concepts have grown more than one name over time — the same entity shows up as one word in the UI, a different word in older screens, and a third in the API or code. That drift confuses users and the AI assistant (which reads the API names).

Each entry below names the canonical term (use this everywhere — UI labels, docs, support, MCP tool descriptions), the synonyms you may still encounter, and the internal / API name so engineers can map the two. When a term is dangerously overloaded, the entry calls out what not to confuse it with.

For contributors

When you add a screen, a doc page, an API field, or an MCP tool, use the canonical term. If you find an old synonym in the product, prefer fixing it over copying it.

The frontend enforces this glossary automatically: msg-zen-test-ai-frontend/docs/Glossary.md is derived from this page, and guard specs fail the frontend build when a forbidden synonym appears in a UI string.

Core entities

Canonical termWhat it isSynonyms you may seeInternal / API name
Test CaseOne natural-language test: header fields, an ordered list of steps, parameters and variants."test", "test definition"TestDefinition
StepA single natural-language instruction inside a test case.TestStep
ActivityA named, contiguous group of steps inside one test case, optionally carrying loop / condition / disabled metadata.activity (TestStep.activityId)
Execution PlanA runnable bundle that organizes test cases into groups, controls their run order, and passes parameter values between them. Can be scheduled or triggered from CI/CD."test set", "test plan"TestSet
GroupAn execution group inside an execution plan — the unit that controls parallel/sequential order.TestSetGroup
FolderThe hierarchy that organizes the list pages (Test-Cases, Snippets, Macros, Execution-Plans). Unrelated to an execution plan's groups.TestGroup (the groupId field)
SnippetA reusable sequence of steps referenced from other test cases.TestDefinition (type SNIPPET)
MacroA small reusable function (e.g. date arithmetic) usable inside steps.macro
ScriptA JavaScript function defined in the tenant and referenced from a step's Scripts setting, used for custom data generation or assertions."script function"TestScriptDefinition
TagA free-form label attached to a test case, used for filtering lists and for driving tag-driven Groups (see above) inside an execution plan."label"Tag
ExecutionOne actual run of a test case or execution plan, with its results, screenshots and logs."run"TestExecution

Parameters & variants

Canonical termWhat it isSynonyms you may seeInternal / API name
ParameterA named placeholder ([[name]]) in step text, URL, etc. whose value is supplied at run time.parameter
VariantOne concrete set of parameter values a test case can run with (plus optional per-run overrides)."parameter combination", "combination"ParameterCombination (combinationId)
note

"Variant" is the canonical word; combinationId survives as the API field name for the same thing. When you read combinationId, think variant.

People, agents & AI — the big overload

The word "agent" is the most overloaded term in the product. It means two completely different things, and they must never be conflated.

Canonical termWhat it isSynonyms you may seeInternal / API name
Test UserThe browser profile / identity a test runs as: auth mode, viewport, language, timezone, entitlements, credentials."user", "agent" (legacy), "browser role"Agent (the agentId field)
AI AgentOne of the AI roles that drives a test: Step-Agent, Login-Agent, XPath-Agent, Assertions-Agent, the Agentic AI, the Correction Agent.(prompt roles in the runner)
AI AssistantThe in-app chat that creates, fixes and reasons about a test over a persistent, resumable session.agent session
Do not confuse

A Test User is who the test logs in as. An AI Agent is the AI that decides what to click. They are unrelated. The entity behind "Test User" is still named Agent in the code and the API (agentId), which is the historical source of the confusion — surface it to humans as Test User everywhere, and reserve the bare word "agent" for the AI roles.

Tenancy & configuration

Canonical termWhat it isSynonyms you may seeInternal / API name
TenantAn isolated tenant: its own tests, users, applications, settings and data."product" (it appears as Product in the API)Product
ApplicationA configured target system: base host/path, login behaviour, proxy, AI hints, iframe handling, etc."host", "enhanced application settings", "ProductHost"ProductHost
Application MapA visual graph of an application's screens (surfaces) and the flows between them, produced by an AI scan."app map"AppMap
SurfaceOne node in an Application Map — a page, step in a flow, or a modal that the scanning agent reached.AppMapNode
note

Product is the internal / API name for Tenant; the /tenant/... URL segment matches the canonical term. ProductHost is an internal schema name for Application — do not expose either internal name to users.

Quality signals

Canonical termWhat it isSynonyms you may seeInternal / API name
FlakinessHow often a test flips between passed and failed on the same, unchanged definition (intermittency), measured as a flip-rate (0–1) over its recent verdicts and shown as Stable / Flaky / Very flaky. It measures trustworthiness, not pass/fail health."flaky test", "intermittent test", "unstable test"FlakinessDto (flip-rate)
note

Always spell it flakiness (one word). A Stable verdict means consistent, not healthy: a test that fails every single time also scores 0 flakiness (it never flips). The stability strip flags that case separately, so "Stable" is never mistaken for "passing".

Quick "use this, not that"

Use thisNot this
Flakinessflaky test · intermittent test · flakyness (misspelling)
Test Casetest definition
Execution Plantest set · test plan
Variantcombination · parameter combination
Test Useragent (when you mean the login identity)
AI Agentagent (only the AI roles are "agents")
Applicationhost · ProductHost · application settings
Application Mapapp map · Anwendungskarte (DE) · Anwendungs-Map (DE)
Scriptscript function
Tenantproduct · Produkt (DE)
Foldergroup (folders organize lists; groups live inside an execution plan)