Skip to content

CONTRIBUTING.md

Latest commit

 

History

History
133 lines (89 loc) · 3.02 KB

CONTRIBUTING.md

File metadata and controls

133 lines (89 loc) · 3.02 KB

Contributing to MalPromptSentinel

Thank you for your interest in contributing to MalPromptSentinel (MPS)!

Contribution Process

MPS uses a contact-first contribution model.

Before making any changes or submitting contributions, please contact the maintainer:

Email: StrategicPromptArchitect@gmail.com

Subject Line: MPS Contribution - [Brief Description]

Include:

  1. What you want to change or add
  2. Why it would improve MPS
  3. Your relevant experience
  4. Proposed approach

Why Contact First?

  1. Coordination - Avoid duplicate work
  2. Architecture - Ensure changes fit the design
  3. Quality - Discuss approach before implementation
  4. Focus - Keep MPS aligned with its goals

Types of Contributions Welcome

High Priority

  • New injection patterns - Real-world attack patterns you've encountered
  • False positive fixes - Benign content incorrectly flagged
  • Bug fixes - Issues in scanning logic
  • Performance improvements - Faster scanning, lower memory

Medium Priority

  • Documentation improvements - Clearer explanations, examples
  • Test cases - New baseline, evasion, or benign tests
  • Integration examples - How you're using MPS

Lower Priority (Discuss First)

  • Architecture changes - Major refactoring
  • New features - Additional capabilities
  • Dependency additions - External libraries

Code Style

If your contribution is approved:

Python

  • Follow PEP 8 style guide
  • Use meaningful variable names
  • Comment complex regex patterns
  • Include docstrings for functions

Patterns

  • Add to mps_patterns.py (not scanner files)
  • Include comment explaining what pattern catches
  • Test against both attack and benign content
  • Document false positive potential

Example Pattern Addition

# In SHARED_PATTERNS dictionary:
'new_category': [
    # Catches: "malicious phrase here"
    # False positives: May match in security documentation
    r'\bmalicious\s+phrase\s+here\b',
],

Testing Requirements

All contributions must:

  1. Pass existing tests

    python -B run_tests.py

  2. Not decrease benign accuracy (currently 93.3%)

  3. Include test cases for new patterns

  4. Document expected impact on detection rates

What Happens After Contact

  1. Review - I'll review your proposal (typically within 1 week)
  2. Discussion - We'll discuss approach if needed
  3. Approval - You'll get go-ahead to implement
  4. Submission - Send completed code via email
  5. Integration - I'll review, test, and integrate
  6. Credit - You'll be acknowledged in CHANGELOG

Code of Conduct

  • Be respectful and professional
  • Focus on technical merit
  • Accept feedback gracefully
  • Give credit where due

Questions?

Email: StrategicPromptArchitect@gmail.com

Website: https://StrategicPromptArchitect.ca

License

By contributing, you agree that your contributions will be licensed under the MIT License.

Thank you for helping improve AI security!