About this project

This is a public, free, open-source demonstration of how to plan an emergency care system, from higher-level use cases to lower-level test automation.

Synthetic data only. All patient names, staff names, vitals, timestamps, and metrics anywhere in this project โ€” the app, the specs, and the tests โ€” are fictitious and generated by Claude AI, for educational purposes. This is not a real emergency care system and was not built from any real hospital's data or workflows.

Specification-driven

Every feature starts as a Gherkin .feature file in tests-with-given-when-then-features/ โ€” the single source of truth that every app screen and every test suite is built to match.

The app

src/ is a real SvelteKit 5 app implementing all 22 features as a single-page dashboard, styled with Lily Design System and SVAR Svelte components. There's no real backend โ€” all data is seeded at load time and, where a scenario needs it to survive a page reload (such as duplicate-patient detection), persisted to localStorage only, in that one browser. Sign in with any free-text identity โ€” a role phrase like "a registration clerk" or a name like "Dr. Smith" โ€” or with doctor@example.com, nurse@example.com, or administrator@example.com (password secret).

The tests

The same 171 scenarios are ported one-to-one across twelve suites, pairing two automation libraries (Selenium, Playwright) with six languages (JavaScript, TypeScript, Python, C#, Java, Rust). See Testing for details.

Skills

skills/ in the source repository packages some of the conventions this project follows (Selenium/JavaScript test-writing, Lily Design System usage) as reusable Claude Code skills.

Everything is in one repository: https://github.com/emergency-care-system/emergency-care-system.