diff --git a/.github/copilot-instructions.md b/.github/copilot-instructions.md
index 71cb68a..5fb3bb9 100644
--- a/.github/copilot-instructions.md
+++ b/.github/copilot-instructions.md
@@ -1,21 +1,206 @@
-Python project for creating Excel xlsx files with openpyxl library.
+# Dienstplan Generator - NRW (Variante 2 "streng")
+
+This is a Python project for automatically generating duty rosters (Dienstplan) with payroll calculations according to NRW (North Rhine-Westphalia, Germany) rules, specifically "Variante 2" which is the strict variant.
+
+## Project Overview
+
+**Purpose**: Automate creation of monthly duty rosters with automatic payroll calculations for weekend/holiday shifts with specific threshold-based compensation rules.
+
+**Language**: Python 3.12+
+**Key Library**: openpyxl (for Excel file manipulation)
+**Output**: Excel workbooks (.xlsx) with multiple sheets including duty plans, rules, holidays, and automated payroll calculations
 
 ## Project Structure
-- `src/main.py` - Main script for creating Excel files
-- `output/` - Directory for generated Excel files
-- `requirements.txt` - Python dependencies (openpyxl)
-- `.venv/` - Python virtual environment
+
+```
+.
+├── src/
+│   ├── build_template.py   # Creates the base Excel template with all formulas
+│   ├── fill_plan_dates.py  # Fills the template with dates for a specific month
+│   ├── read_excel.py       # Utility for reading Excel files
+│   └── main.py             # Legacy demo script (not main entry point)
+├── output/                 # Generated monthly duty rosters (gitignored)
+├── templates/              # Base template files (gitignored, generated)
+├── requirements.txt        # Python dependencies (openpyxl==3.1.2)
+├── SPECIFICATION.md        # Complete business rules and Excel formulas (German)
+└── README.md              # User documentation (German)
+```
 
 ## Setup Instructions
-1. Virtual environment is already configured
-2. Dependencies are already installed
-3. Run the script: `python src/main.py`
+
+1. **Create virtual environment** (if not exists):
+   ```bash
+   python -m venv .venv
+   ```
+
+2. **Activate virtual environment**:
+   - Windows PowerShell: `.\.venv\Scripts\Activate.ps1`
+   - Linux/Mac: `source .venv/bin/activate`
+
+3. **Install dependencies**:
+   ```bash
+   pip install -r requirements.txt
+   ```
 
 ## Usage
-The main script demonstrates how to:
-- Create Excel workbooks
-- Add and format headers
-- Insert data rows
-- Apply styling (fonts, colors, alignment)
-- Adjust column widths
-- Save files with timestamps
+
+### Creating Monthly Duty Rosters
+
+1. **Build template** (first time or after template changes):
+   ```bash
+   python src/build_template.py
+   ```
+   This creates `templates/Dienstplan_Vorlage_V2_NRW.xlsx` with all formulas
+
+2. **Generate a specific month**:
+   ```bash
+   python src/fill_plan_dates.py 2025 11  # November 2025
+   python src/fill_plan_dates.py 2025 12  # December 2025
+   ```
+   This creates `output/Dienstplan_YYYY_MM_NRW.xlsx` pre-filled with all dates
+
+3. **Manual data entry in Excel**:
+   - Open the generated file
+   - Go to "Plan" sheet
+   - Enter employee names in column B
+   - Enter work share (Anteil) in column C (0.5 = half shift, 1.0 = full shift)
+   - Go to "Auswertung" sheet and list all employees in column A
+   - All calculations happen automatically
+
+## Domain-Specific Context
+
+### Business Rules (Variante 2 - "streng")
+
+**WE-Tag** (Weekend/Holiday shift) includes:
+- Friday, Saturday, Sunday
+- Public holidays (Bundesland-specific, currently NRW)
+- Day before a public holiday ("Vortag")
+
+**WT-Tag** (Weekday shift):
+- All days that are NOT WE-Tag
+- Always compensated at 250€ per unit
+
+**WE Compensation (strict variant)**:
+- Only paid if monthly total ≥ 2.0 WE units (threshold)
+- If threshold reached: 450€ per WE unit
+- After reaching threshold: subtract exactly 1.0 WE unit (priority: Friday first, then other WE days)
+- Below threshold: 0€ for WE shifts (NOT converted to WT compensation)
+
+### Excel Structure
+
+The generated Excel workbooks contain these sheets:
+
+1. **README**: Quick reference for users
+2. **Regeln**: Parameters (rates, threshold, Bundesland selection, month selection)
+3. **Feiertage**: Table of public holidays (tblFeiertage) with Date, Name, BL columns
+4. **Plan**: Main data entry (tblPlan) - Date, Employee (Mitarbeiter), Share (Anteil)
+5. **Auswertung**: Automatic payroll calculations per employee per month
+6. **Checks**: Quality checks (e.g., daily shift shares should sum to 1.0)
+
+### Key German Terms
+
+- **Dienstplan** = Duty roster
+- **Mitarbeiter** = Employee
+- **Anteil** = Share/portion (0.5 = half shift, 1.0 = full shift)
+- **WE** = Wochenende (Weekend/Holiday)
+- **WT** = Werktag (Weekday)
+- **Feiertag** = Public holiday
+- **Vortag** = Day before
+- **Schwelle** = Threshold
+- **Abzug** = Deduction
+- **Auszahlung** = Payout
+- **Bundesland (BL)** = German federal state
+- **NRW** = Nordrhein-Westfalen (North Rhine-Westphalia)
+
+## Code Style
+
+- Use German variable names in Excel formulas (matches SPECIFICATION.md)
+- Use English in Python code with German terms in comments where needed
+- Follow PEP 8 for Python code
+- Use descriptive variable names
+- Add docstrings to functions explaining business logic
+- Comment complex Excel formula generation with business rule references
+
+## Excel Formula Guidelines
+
+- Use German Excel function names (e.g., `SUMMENPRODUKT`, `WENN`, `MONATSENDE`)
+- Use semicolon `;` as argument separator (German Excel convention)
+- Use comma `,` as decimal separator in Excel
+- Prefer `SUMMENPRODUKT` over Office 365-specific functions for compatibility
+- Always handle empty cells with `WENN` guards to avoid errors
+- Use tolerance of 0.0001 for threshold comparisons (floating-point safety)
+
+## Testing
+
+Currently, the project has no automated tests. Manual testing involves:
+
+1. **Template generation test**:
+   ```bash
+   python src/build_template.py
+   # Check: templates/Dienstplan_Vorlage_V2_NRW.xlsx exists and opens
+   ```
+
+2. **Month generation test**:
+   ```bash
+   python src/fill_plan_dates.py 2025 11
+   # Check: output/Dienstplan_2025_11_NRW.xlsx exists with November dates
+   ```
+
+3. **Business logic tests** (manual in Excel):
+   - Enter test data according to test cases in SPECIFICATION.md
+   - Verify calculations match expected results
+   - Test edge cases: under threshold (< 2.0), exactly at threshold (2.0), over threshold
+
+## Common Tasks
+
+### Adding a New Holiday
+
+Edit `src/build_template.py`:
+- Add to `NRW_HOLIDAYS_2025` or `NRW_HOLIDAYS_2026` lists
+- Rebuild template: `python src/build_template.py`
+
+### Changing Payroll Rules
+
+Edit `src/build_template.py` in `_populate_rules()`:
+- Modify Satz_WT, Satz_WE, WE_Schwelle, or Abzug_nach_WE_Schwelle defaults
+- Rebuild template: `python src/build_template.py`
+
+### Supporting a New Bundesland
+
+1. Add holiday data in `src/build_template.py`
+2. Ensure BL_Auswahl dropdown includes the new state
+3. The formulas already filter holidays by BL automatically
+
+## Known Limitations
+
+- No automated tests
+- Template must be regenerated after code changes (not dynamic)
+- Only NRW holidays pre-populated (2025-2026)
+- Excel must support tables and formulas (no LibreOffice/Google Sheets tested)
+- German Excel required (function names, argument separators)
+
+## Important Files
+
+- **SPECIFICATION.md**: Complete source of truth for business rules and Excel formulas (in German)
+- **README.md**: User-facing documentation (in German)
+- **src/build_template.py**: Main implementation of Excel template with all formulas
+- **src/fill_plan_dates.py**: CLI tool for generating monthly files
+
+## When Making Changes
+
+1. **Business rule changes**: Update SPECIFICATION.md first, then implement in build_template.py
+2. **Formula changes**: Test in Excel manually with edge cases before committing
+3. **Python changes**: Ensure backwards compatibility with existing templates
+4. **Always**: Regenerate template and test with a sample month after changes
+
+## Dependencies
+
+- Python 3.12+ (developed with 3.12.3)
+- openpyxl 3.1.2 (Excel file manipulation)
+- No other external dependencies
+
+## File Encoding
+
+- Python files: UTF-8
+- Excel files: Binary .xlsx format
+- All text content supports German umlauts (ä, ö, ü, ß)