--- title: "Key Concepts and Glossary" slug: "key-concepts" description: "Plain definitions of the terms used across Validatar and its docs, from catalog and child test to trust score and Viewer, each linked to its main page." status: "new" updated: 2026-11-20T21:54:23Z published: 2026-11-20T21:49:30Z canonical: "docs.validatar.com/key-concepts" --- > ## Documentation Index > Fetch the complete documentation index at: https://docs.validatar.com/llms.txt > Use this file to discover all available pages before exploring further. # Key Concepts and Glossary This glossary defines the terms you'll meet across Validatar and these docs, in alphabetical order. Each entry links to the page that covers it in full. Examples use the data from the rest of the docs: the **Wholesale Distribution System** (SQL Server) and the **Enterprise Data Warehouse** (Snowflake) it loads into. ## Catalog The part of Validatar, under **Catalog** in the header, that lists every data source, schema, table and column Validatar has ingested. Each table and column page shows its metadata, profiles, the tests that check it, lineage and its trust score. See [Catalog Overview](/docs/catalog-overview). ## Child test One of the tests a template test creates, one per selected table or column. A child test runs and reports on its own, so a failure names its table. A child test nobody has edited is non-customized and can be updated from the template when you materialize again; one edited by hand is customized. See [Template Tests](/docs/template-tests). ## Creator A user type that can create, edit and import tests. Each Creator uses one of the Creator seats in your license. Users who aren't Creators are Viewers. See [Validatar Roles](/docs/validatar-role-help). ## Data Agent A service you install on a Windows or Linux machine, or in Kubernetes, inside your network. It runs queries and Python scripts for Validatar against sources Validatar can't reach directly, such as a SQL Server behind a firewall. See [Validatar Data Agent](/docs/validatar-data-agent). ## Data processing engine The engine that compares the results of a test's two data sets and decides pass or fail. **Built-In** compares them inside Validatar; **Snowflake** and **Databricks** compare them inside your own account, so only the summary comes back. See [What Is a Data Processing Engine?](/docs/what-is-a-data-processing-engine) ## Data source Validatar's connection to one system that holds data: a database, a file location, an API or a report service. Tests, profile sets and the Catalog all work through data sources, and each data source has its own permissions. See [Create a Data Source](/docs/create-a-data-source). ## Data source template A reusable definition for one platform, such as SQL Server or Snowflake. It tells Validatar how to ingest that platform's metadata, which profiles it supports, how to quote names and how to map data types. Every data source is built from one. See [What Is a Data Source Template?](/docs/what-is-a-data-source-template-overview) ## Folder A way to organize tests and jobs inside a project. Each project starts with a **Tests** folder and a **Jobs** folder, and you add subfolders under them. A job can run a whole folder as one step. See [Organizing Projects](/docs/organizing-projects). ## Impact rating How much a table matters to the business, from **1 - Very Low** to **5 - Critical**, plus **0 - Unknown**. The rating sets the trust score the table should reach, which decides whether it shows **Above Target** or **Below Target**. See [Impact Ratings](/docs/impact-ratings). ## Job A set of steps that run together: tests, whole folders, profile sets and other jobs. A job step can materialize a template's child tests before running them. Jobs run on demand or on a schedule. See [Create a Job](/docs/job-details). ## Label A tag you put on tests, template tests and jobs to group them across folders, for example everything that belongs to one release. Labels help you find a set of tests and export it together. See [Labels Overview](/docs/labels-overview). ## Marketplace The section under **Marketplace** in the header where you browse content published by Validatar, such as test packs, data source templates, reports, custom fields and Agent Kits, and import it into your instance. See [Navigating Validatar](/docs/navigating-validatar). ## Materialize The step that turns a template test into child tests. Materializing creates a child test for each newly selected object and, depending on the options you choose, updates non-customized child tests and deletes child tests for objects no longer selected. Nothing runs until a template is materialized. See [Materialize Child Tests](/docs/materializing-child-tests). ## MCP server Validatar's Model Context Protocol endpoint, which lets outside AI tools such as Claude Code, GitHub Copilot, Cursor and Codex use the same tools as Validatar AI. It needs an MCP token and works on Validatar Cloud and Validatar Server from 2026.1. See [Connecting to the Validatar MCP Server](/docs/connecting-to-the-validatar-mcp-server). ## Metadata Information about your data rather than the data itself. Ingested metadata is the structure Validatar reads from a data source: schemas, tables, columns and data types. Custom metadata is what people add, such as descriptions, owners and custom field values. Tests and templates can use both. See [Ingesting Schema Metadata](/docs/ingesting-schema-metadata). ## Metadata link A link from a test's data set to a table or column in the Catalog that the test checks. Linked tests appear on that table's or column's **Tests** tab, and links marked **Impacts Trust Score** count toward its trust score. See [Understanding Trust Scores](/docs/understanding-trust-scores). ## Profile and profile set A profile is one statistic about a table or column, such as its row count, null count or number of distinct values. A profile set chooses which profiles to collect for which objects and runs on demand, in a job or on a schedule. Results show on the **Profile History** tab in the Catalog. See [Data Profiles](/docs/data-profiles) and [Create and Run a Profile Set](/docs/create-and-run-a-profile-set). ## Project The container for a team's tests, jobs, results and schedules. Each project has its own members and permissions and its own list of data sources it may test. In the example docs, **Warehouse Reconciliation** is the project that checks the warehouse against the Wholesale Distribution System. See [Create a Project](/docs/create-a-project). ## Quality dimension The kind of quality a test checks, such as Completeness, Uniqueness, Validity, Consistency, Accuracy or Timeliness. Each test has one. The number of dimensions a table's tests cover drives its coverage score. See [Quality Dimensions](/docs/quality-dimensions). ## Quick test A ready-made check you add from a table's or column's **Tests** tab in the Catalog. **Add Test** opens the Test Wizard ("What can we check for you?"). A quick test is a template test underneath, so one quick test can cover many columns at once. See [Add Quick Tests from the Catalog](/docs/add-tests-from-the-catalog). ## Schedule The times a job runs, such as every day at 5:00 AM or the second Friday of each month, and who gets the results by email. See [Schedule a Job](/docs/scheduling-a-job). ## Severity How serious a test failure is: **Critical**, **High**, **Medium**, **Low** or **Informational** by default. Each severity has a weight in the quality score; out of the box a Critical run counts 16 times as much as a Low one, and Informational counts for nothing. See [Severity Levels](/docs/severity-levels). ## Standard test One specific check: a data set (a query or script), what to compare it to (another data set, a rule or earlier runs), and the result criteria that decide pass or fail. **Wholesale Invoiced Sales Tie Out by Month** is a standard test. See [Create a Standard Test](/docs/create-a-standard-test). ## Template test One test definition applied to many tables or columns. The object list comes from the Catalog (**Metadata Driven**), a query (**Metadata Script**) or a spreadsheet (**Import File**), and Validatar creates a child test for each object. **Wholesale Row Counts Match the Warehouse** covers 49 tables with one template. See [Template Tests](/docs/template-tests). ## Test action A follow-up that runs after a standard test, when the test runs, fails or changes status. A step can send an email, run another test, run a script, update metadata or open a Jira issue. See [What Are Test Actions](/docs/what-are-test-actions). ## Trust score A score from 0 to 5 stars for each table, calculated once a day from the runs of the tests linked to it over the look back period (14 days by default). It multiplies a quality score, weighted by severity, by a coverage score based on quality dimensions. See [Understanding Trust Scores](/docs/understanding-trust-scores). ## Validatar AI The AI chat panel in Validatar, opened from the Validatar AI icon in the header, plus inline actions such as **Auto-Document** and **Recommend Table Tests**. It can create tests and data sources and run standard tests, and every change can be undone from the chat history. It's on Validatar Cloud as a beta; contact sales to enable it. See [Using the Validatar AI Chat Interface](/docs/using-the-validatar-ai-chat-interface). ## Viewer The user type for everyone who isn't a Creator. Viewers see the projects, data sources and reports their permissions allow and can run tests and jobs where permitted, but they can't create, edit or import tests. See [Validatar Roles](/docs/validatar-role-help). ## Related - [What Is Validatar?](/what-is-validatar.md) - [How Validatar Works](/test-automation-lifecycle.md) - [Navigating Validatar](/navigating-validatar.md) - [Validatar Roles](/validatar-role-help.md)