CONTRIBUTING.md

Contributing to MemMark

Thank you for your interest in contributing to MemMark! This document provides guidelines and instructions for contributing.

Development Setup

1. Clone the repository:

   git clone https://github.com/Carlos-Projects/memmark.git
   cd memmark

2. Create a virtual environment:

   python -m venv .venv
   source .venv/bin/activate  # On Windows: .venv\Scripts\activate

3. Install dependencies:

   pip install -e ".[dev]"

Code Style

• Follow PEP 8 conventions
• Use type hints on all functions and methods
• Maximum line length: 88 characters
• Use ruff for linting:

  ruff check src/ tests/

Testing

• Write tests for all new functionality
• Minimum coverage: 80%
• Run tests with:

  python -m pytest tests/ -v

Pull Request Process

1. Create a feature branch from main
2. Make your changes with appropriate tests
3. Ensure all tests pass and coverage meets the threshold
4. Run ruff check and fix any issues
5. Submit a pull request with a clear description of changes

Commit Messages

Follow conventional commits:

• feat: New feature
• fix: Bug fix
• docs: Documentation changes
• test: Test additions or modifications
• refactor: Code refactoring
• chore: Maintenance tasks

Architecture

MemMark follows a modular architecture:

src/memmark/
├── scanner.py          # Core scanning engine
├── cli.py              # Typer CLI interface
├── watermark/          # Memory watermarking
├── poisoning/          # Poisoning detection
├── provenance/         # Provenance tracking
├── integrity/          # Integrity verification
├── reporters/          # Output formatters
└── utils/              # Shared utilities

Integration with Ecosystem

• Uses mcp-taxonomy for standardized finding classification
• Generates policies compatible with MCPGuard
• Reports consumable by MCPscop
• Complements reverse-abliterate (model integrity vs memory integrity)

Questions?

Open an issue on GitHub or reach out via the project discussions.