What problem does it solve? Writing integration and E2E tests usually fails because tests are authored against guessed selectors, invented URLs, or assumed environments. This Skill eliminates that guesswork by requiring the agent to explore the running application first, reuse a shared test environment, and follow the repository's existing test conventions. ## Core Features & Use Cases - Live-app exploration before authoring: Drives the configured browser provider (open, snapshot, interact, assert) to capture real selectors, flows, and validation messages before writing any test. - Shared test environment reuse: Attaches to the om-prepare-test-env descriptor at .ai/qa/test-env.json after validating PID, readiness probe, and freshness, avoiding duplicate environment bootstraps. - Repository-native conventions: Discovers existing runners (Playwright, Cypress, WebdriverIO), mirrors test layout and naming, and keeps timeouts/retries in the shared runner config. - Artifact-based failure analysis: Classifies every failure as product regression, test issue, or environment issue, with a mandatory per-test evidence table and ownership assignment. - Use Case: After implementing a company-creation flow, ask the agent to write integration tests; it attaches to the running test environment, walks the flow in the browser, authors a Playwright test using observed roles and labels, runs it, and reports results with evidence. ## Quick Start Ask the agent to write and run integration tests for the login flow using the om-integration-tests skill against the shared test environment.
