← ClaudeAtlas

nanook-app-runnerlisted

Connects an application to a table-driven Playwright test execution: one shared runner over generated suite files, page objects under a common contract, base states created via API (login optimization), and resource pooling by effect profile. Execution side — table authoring is create-equivalence-class-table. Trigger: "connect app to nanook", "table-driven test execution", "page object contract", "base state setup", "suite runner", "user pooling", "nanook app runner"
xhubio/nanook-skill · ★ 1 · Data & Documents · score 62
Install: claude install-skill xhubio/nanook-skill
# Nanook App Runner — Table-Driven Test Execution for an Application This skill describes the **execution side** of table-driven testing: how the test cases that Nanook tables generate become a running Playwright (or API) test suite for an application — without hand-written spec files per case. | Question | Skill | |---|---| | How do I write the tables (EqClasses, CASCADE, 100%, references)? | `create-equivalence-class-table` | | How do the tables become a running test execution for an app? | **this one** | ## The Architecture in One Picture ``` Table (.xlsx, source of truth) │ build step (nanook-table + a suite writer) ▼ suites/*.json ← NO Playwright vocabulary (deliberately) │ case = base state + actions + expectation (ui/api) ├────────────► API runner checks the API boundary ▼ Browser runner (Playwright) checks the UI boundary │ calls per app only a small "connection" object ▼ <app>/tests/suites.spec.ts thin layer: ONE call into the shared runner ``` The division of labor — each row is a hard boundary: | | who | why | |---|---|---| | Which cases exist | **table** | the domain opinion, editable without TypeScript | | What the page can do and shows | **page object** | abstracts the UI, **never judges** | | Whether it is correct | **shared runner** | the expectation stays in ONE place | | What the app does differently | **app connection** | base state + entry point — and ONLY those | |