diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
new file mode 100644
index 0000000..4ab0199
--- /dev/null
+++ b/CONTRIBUTING.md
@@ -0,0 +1,388 @@
+# Contributing to SynthOps
+
+Thank you for your interest in contributing to SynthOps.
+
+SynthOps is an open-source synthetic operational data generation framework designed to create realistic, structured and fictional datasets for analytics engineering, business intelligence, AI prototyping and decision-intelligence use cases.
+
+The project is currently in early development. Contributions are welcome, especially around documentation, testing, domain modelling, data quality rules, examples and new synthetic data generation modules.
+
+---
+
+## Project Goals
+
+SynthOps aims to:
+
+- generate safe fictional datasets for analytics and prototyping
+- support multiple operational domains through modular design
+- keep the core engine domain-neutral
+- keep domain-specific logic modular
+- document synthetic assumptions clearly
+- maintain realistic relationships between generated tables
+- support reproducible data generation
+- provide examples useful to analysts, analytics engineers, BI developers and data scientists
+
+---
+
+## Current Project Status
+
+SynthOps is in early development.
+
+The first domain module is:
+
+```text
+adult_social_care
+```
+
+Currently implemented tables:
+
+```text
+care_homes
+residents
+```
+
+Planned Adult Social Care tables include:
+
+```text
+resident_care_needs_history
+staff
+shifts
+incidents
+observations
+handover_notes
+```
+
+Future planned domains may include:
+
+```text
+finance_operations
+construction_operations
+saas_metrics
+workforce_analytics
+```
+
+---
+
+## Ways to Contribute
+
+You can contribute by helping with:
+
+- improving documentation
+- adding tests
+- reviewing data generation assumptions
+- improving sample outputs
+- suggesting realistic operational scenarios
+- fixing bugs
+- improving developer experience
+- adding new domain modules
+- adding example notebooks or dashboards
+- improving validation rules
+- improving diagrams and data dictionaries
+
+---
+
+## Responsible Synthetic Data Principles
+
+SynthOps must not use real personal, clinical, financial, employment or organisational records.
+
+All generated data should be fictional.
+
+When contributing, do not add:
+
+- real resident, patient, staff or provider records
+- real NHS numbers or health identifiers
+- real addresses linked to individuals
+- real phone numbers or emails belonging to individuals
+- real financial account details
+- real employment records
+- real safeguarding records
+- real clinical records
+- real commercially confidential datasets
+
+Synthetic assumptions should be documented clearly and should not be presented as official benchmarks or real-world statistics.
+
+---
+
+## Development Setup
+
+Clone the repository:
+
+```powershell
+git clone https://github.com/Phoenix-maho/synthops.git
+cd synthops
+```
+
+Create a virtual environment:
+
+```powershell
+python -m venv .venv
+```
+
+Activate the virtual environment on Windows PowerShell:
+
+```powershell
+.\.venv\Scripts\Activate.ps1
+```
+
+Install the project in editable mode with development dependencies:
+
+```powershell
+pip install -e ".[dev]"
+```
+
+Run the tests:
+
+```powershell
+pytest
+```
+
+---
+
+## Branching Workflow
+
+Create a new branch for each meaningful change.
+
+Use descriptive branch names, such as:
+
+```text
+feature/resident-care-needs-history
+docs/update-data-dictionary
+test/care-home-validation
+fix/custom-id-format
+ci/add-test-workflow
+```
+
+Example:
+
+```powershell
+git checkout -b feature/resident-care-needs-history
+```
+
+Avoid making feature changes directly on `main`.
+
+---
+
+## Commit Guidelines
+
+Use clear commit messages that describe the change.
+
+Good examples:
+
+```text
+Add resident care needs history generator
+Document adult social care data dictionary
+Fix custom ID width validation
+Add tests for resident lifecycle rules
+```
+
+Avoid vague commit messages such as:
+
+```text
+update
+fix stuff
+changes
+work
+```
+
+A good commit should represent one meaningful unit of work.
+
+---
+
+## Pull Request Guidelines
+
+Before opening a pull request:
+
+- make sure tests pass
+- update documentation if behaviour changed
+- add tests for new generator logic
+- keep changes focused
+- explain the purpose of the change
+- mention assumptions introduced or changed
+- link the relevant issue using a closing keyword where appropriate
+
+A good pull request should answer:
+
+1. What changed?
+2. Why was the change needed?
+3. How was it tested?
+4. Are there any limitations or assumptions?
+
+Example closing keyword:
+
+```text
+Closes #12
+```
+
+Make sure closing keywords are outside code blocks so GitHub can detect them.
+
+---
+
+## Adding a New Domain Module
+
+New domain modules should follow the existing structure:
+
+```text
+src/synthops/domains/new_domain/
+├── __init__.py
+├── table_generator.py
+└── generate.py
+```
+
+Each domain should include:
+
+- source code
+- tests
+- documentation
+- sample output
+- example usage
+- documented assumptions
+- relevant data dictionary updates
+
+Domain-specific logic should stay inside the domain module.
+
+Reusable logic should go into:
+
+```text
+src/synthops/core/
+```
+
+---
+
+## Adding a New Generator Table
+
+When adding a new table generator, include:
+
+- generator function
+- configurable ID prefix and width where relevant
+- random seed support
+- realistic value ranges
+- referential integrity rules
+- validation or tests for key assumptions
+- sample output where appropriate
+- documentation updates
+- changelog entry
+
+Example expected pattern:
+
+```python
+def generate_example_table(
+    number_of_rows: int = 10,
+    id_prefix: str = "EX",
+    id_width: int = 4,
+    seed: int | None = None,
+):
+    ...
+```
+
+---
+
+## Testing Expectations
+
+All important generator behaviour should be tested.
+
+Tests should cover:
+
+- expected output type
+- expected row counts
+- expected columns
+- unique IDs
+- custom ID formats
+- invalid inputs
+- valid value ranges
+- date consistency
+- referential integrity
+- reproducibility when using seeds
+
+Run tests with:
+
+```powershell
+pytest
+```
+
+GitHub Actions also runs the test suite automatically on pull requests and pushes to `main`.
+
+---
+
+## Documentation Expectations
+
+Documentation should be updated when:
+
+- a new feature is added
+- a new domain module is introduced
+- a generated table changes
+- assumptions change
+- responsible-use guidance changes
+- setup or installation steps change
+- sample outputs change
+
+Documentation should be clear enough for:
+
+- analysts
+- analytics engineers
+- BI developers
+- data scientists
+- open-source contributors
+- technical reviewers
+
+---
+
+## Code Style
+
+Current style expectations:
+
+- write clear, readable Python
+- prefer descriptive function and variable names
+- keep domain assumptions visible
+- avoid unnecessary cleverness
+- separate reusable core logic from domain-specific logic
+- include docstrings for public generator functions
+- keep functions focused and testable
+
+More formal linting and formatting tools may be added later.
+
+---
+
+## Reporting Issues
+
+When reporting an issue, include:
+
+- what you expected to happen
+- what actually happened
+- steps to reproduce the issue
+- your operating system
+- Python version
+- relevant error messages
+- example input if applicable
+
+---
+
+## Suggesting New Features
+
+Feature suggestions should explain:
+
+- the problem being solved
+- who would benefit
+- the proposed behaviour
+- any relevant domain assumptions
+- whether it belongs in the core engine or a domain module
+
+For new domain suggestions, explain why that domain would benefit from synthetic operational data.
+
+---
+
+## Maintainer Notes
+
+SynthOps is being developed as a professional open-source product.
+
+Changes should improve at least one of the following:
+
+- usefulness
+- reliability
+- realism
+- maintainability
+- documentation
+- developer experience
+- test coverage
+- responsible data practice
+- community adoption
+
+The goal is not simply to add more generated tables. The goal is to build a trustworthy synthetic data product that others can understand, use and extend.
\ No newline at end of file