What problem does it solve? Writing and maintaining Playwright end-to-end tests is error-prone when agents guess selectors, invent API calls, or skip failure evidence. This Skill enforces a staged, rules-driven protocol so tests are discovered from the live app, written against a canonical Steps API, and diagnosed from real failure artifacts instead of guesswork. ## Core Features & Use Cases - Staged test authoring pipeline: Runs scenario discovery, live DOM inspection via @playwright/cli, spec writing, and a three-part review (optimization, API compliance, composition judge) before any test is accepted. - Element-repository discipline: All selectors live in page-repository.json and are referenced by name, with a hard ban on inline selectors in durable specs. - Failure diagnosis routing: Local and CI pipeline failures are dispatched to a subagent-only failure-diagnosis protocol with a mandatory evidence floor (trace, DOM state, console). - Companion skill orchestration: Routes coverage expansion, bug discovery, contract/database/performance testing, and test repair to dedicated companion skills. - Use Case: A user says "the nightly regression failed" — the Skill pins the run's commit, pulls CI artifacts with gh, and dispatches a failure-diagnosis subagent rather than re-running the suite locally. ## Quick Start Ask the agent to write an end-to-end test for your app's checkout flow and it will discover the scenario, inspect the live page, and produce a stabilized Playwright spec.
