Glossary
Plain-English definitions for the terms used throughout this documentation and in the PACE E2E codebase.
| Term | Definition |
|---|---|
| Spec | A Markdown file in tests/specs/ that describes a test scenario in plain English — frontmatter metadata, numbered steps, and assertions. Specs are the input to the Generator agent and the source of truth for what a test should do. |
| Test data | Structured, Zod-validated JSONC files in tests/data/ that provide project IDs, names, and other values the tests need. Referenced in specs and test files via {{env:...}} or the typed data fixture. |
| Page Object Model (POM) | A TypeScript class in tests/page-objects/ that encapsulates the selectors and interactions for one area of the PACE UI. Tests call POM methods instead of working with raw Playwright locators directly. |
| Helper | A utility function in tests/utils/ that handles a specific cross-cutting concern — for example, waiting for an AG-Grid to finish loading, interacting with a form field by label, or switching to a tab. Helpers are called by POMs and tests. |
| Fixture | A Playwright test fixture (files in tests/fixtures/) that sets up shared state before each test — for example, auth.fixture.ts handles login and wires up POMs, while data.fixture.ts exposes typed test data. |
| Agent | One of the four AI agents in the framework — Planner, Generator, Reviewer, or Stabilizer — each with a canonical host-neutral prompt in tests/agents/ and thin per-host adapter files. |
| Manifest | The frontmatter block at the top of a spec file (YAML) that records metadata such as the feature area, data references, and pipeline status. Also used informally to refer to the tests/data/*.jsonc data files. |
| AG-Grid | The third-party data grid library used throughout the PACE application. Test helpers in tests/utils/ include dedicated utilities for scrolling, filtering, and reading cells in AG-Grid instances. |
| Smoke run | A single quick execution of a generated test by the Generator agent to confirm the test does not throw an immediate error. A passing smoke run does not mean the test is fully correct — that is the Reviewer's job. |
| Headed vs headless | Whether the browser window is visible during a test run. Headless (default, HEADLESS=true) runs the browser invisibly in the background — faster and suited for CI. Headed (HEADLESS=false) opens a visible browser window — useful for debugging what the test is doing. |
Environment (TEST_ENV) | The deployment target for the test run, selected by the TEST_ENV environment variable. Each value maps to a different base URL (e.g. dev2 → https://dev2saas.frontrol.com). Per-environment data overrides live in tests/data/env/. |