A fully self-contained, offline-capable, single-file web application for generating and managing EWTD-compliant medical staff rosters. No server, no build pipeline, no internet connection required — just open src/index.html in a modern browser.

1. What the Application Does
2. Prerequisites
3. Installation
4. Running the Application
5. Running the Tests
6. Configuration Options
7. Shift Types
8. EWTD Compliance Rules
9. Supported Regions & Public Holidays
10. Exporting Rosters
11. Data Storage & Privacy

Medical Roster is an EWTD-compliant staff rostering tool designed for healthcare administrators. It provides:

• Automated roster generation — a greedy, constraint-respecting algorithm assigns shifts to staff members across a configurable reference period (up to 17 weeks, up to 50 staff) in under 5 seconds.
• EWTD compliance enforcement — all four Working Time Directive rules are checked automatically. Any violation is surfaced with a precise, human-readable message (e.g. "Minimum 11-hour rest not met between shift A and shift B").
• Monthly calendar view — fully navigable month-by-month calendar with weekends and public holidays clearly distinguished. Roster entries are colour-coded by shift type.
• Staff management — add, edit, and remove staff members with per-member constraints including absence dates, approved leave periods, shift-type restrictions, and contracted hours — all without page reload.
• Rest-day logic — optional automatic insertion of mandatory rest days based on configurable criteria (e.g. post-night-shift recovery, maximum consecutive shifts).
• Three shift configurations — 24-hour rolling, 12-hour day/night split, and sub-divided day portions (AM/PM/evening).
• Export — download the completed roster as CSV or HTML, or print to PDF via the browser print dialog.
• Offline, zero-dependency — all logic, styles, and bundled public holiday data are inlined in a single HTML file. No CDN, no backend, no analytics.

• A modern browser: Google Chrome (current stable), Mozilla Firefox (current stable), or Microsoft Edge (current stable).
• No internet connection, server, or build tools required.

• Node.js v16 or later (the tests use only Node.js built-ins; no npm install step is needed).

No installation is needed to run the application itself.

git clone <repo-url>
# or simply download the repository as a ZIP and extract it

The application works correctly when served from any path, e.g. /tools/roster.html. No root-relative paths are used.

# Example: Python simple server (optional, for hosted deployment)
python3 -m http.server 8080
# Then open http://localhost:8080/src/index.html

1. Navigate to the src/ directory in your file manager (or via the terminal).
2. Open src/index.html directly in Chrome, Firefox, or Edge.

# Linux / macOS
xdg-open src/index.html       # Linux
open src/index.html            # macOS

# Windows
start src\index.html

The application loads immediately. Your staff list and roster configuration are automatically saved to browser localStorage and restored on every subsequent visit — no re-entry required.

An in-page Instructions section (accessible from the top navigation) guides new users through:

1. Selecting a region and reference period.
2. Adding staff members and their constraints.
3. Choosing a shift type.
4. Generating a compliant roster.
5. Reviewing and resolving any EWTD violations.
6. Exporting the finished roster.

All tests are plain Node.js scripts — no test runner or npm install is required.

npm test

This runs the persistence, EWTD compliance engine, and rest-day logic test suites in sequence.

# Persistence / localStorage layer
npm run test:persistence

# EWTD compliance engine (all four rules)
npm run test:ewtd

# Rest-day rostering logic
npm run test:rest-day

# Full roster generation algorithm
npm run test:roster

# Comprehensive EWTD suite
node tests/ewtd_comprehensive_suite.test.js

# localStorage persistence validation
node tests/localstorage_persistence_validation.test.js

# Performance benchmark (50 staff × 17 weeks)
node tests/performance_benchmark.js

Each test script prints a summary to stdout. The performance benchmark additionally reports the elapsed time for roster generation. All tests exit with code 0 on success and 1 on failure.

All configuration is performed within the application UI and persisted automatically. The options below correspond to the settings panels in the app.

Each staff member supports the following fields:

When Rest-Day Logic is enabled:

A single shift covering a full 24-hour period. Configurable start and end times (typically 08:00–08:00 next day).

Two shifts per day:

Up to three named portions within the day shift, for example:

All start and end times for each shift type are configurable in the Shift Settings panel.

The compliance engine (src/ewtd_compliance_engine.js) enforces four rules automatically:

Every violation is reported with:

• The rule ID and description.
• The affected staff member's name.
• The specific date(s) involved.
• A human-readable message detailing the exact breach (e.g. "8.5 h rest recorded between Day 2025-06-10 and Night 2025-06-11; minimum is 11 h").

The roster generation algorithm will either produce a zero-violation roster or explicitly flag every violation — it will never silently ignore a breach.

Holiday data is statically bundled inside the HTML file for years 2023–2028. No network request is ever made.

Public holidays are visually distinguished in the calendar view and are treated as non-working days by default during roster generation.

From the Export panel in the application:

All exports include the full roster data without loss of information.

• All data is stored exclusively in your browser's localStorage for the domain/path from which the file is opened.
• No data is transmitted to any server, third party, analytics service, or telemetry endpoint — ever.
• To clear all stored data, use your browser's developer tools to clear localStorage, or use the Reset option within the application settings panel.
• Staff names and scheduling data never leave the user's own browser.

Built by HermesOrg — an orchestration of distinct AI personas working in concert. Product Manager → Designer → Engineer → QA, each contributing expertise to build something real.