Skip to content

CONTRIBUTING.md

Latest commit

 

History

History
61 lines (44 loc) · 1.92 KB

CONTRIBUTING.md

File metadata and controls

61 lines (44 loc) · 1.92 KB

Contributing to Prompt Guard

Thanks for your interest in contributing! This project is open to anyone who wants to help improve prompt injection detection.

How to contribute

Reporting bugs

Open an issue using the Bug Report template. Include:

  • Steps to reproduce
  • Expected vs actual behavior
  • Python version and OS

Suggesting new patterns

If you've found a prompt injection technique that Prompt Guard doesn't catch:

  1. Open an issue with the Feature Request template
  2. Include example text that should be detected
  3. Suggest the severity level (critical/high/medium/low)

Submitting Pull Requests

  1. Fork the repo
  2. Create a branch from main: git checkout -b my-feature
  3. Make your changes
  4. Test locally: python prompt_guard.py ./test-directory
  5. Open a PR with a clear description of what you changed and why

Adding detection patterns

Patterns live in prompt_guard.py inside the PATTERNS dict. Each entry has:

(r"regex_pattern", "unique_name", "Human-readable description")

Guidelines:

  • Use re.IGNORECASE (applied automatically)
  • Keep regexes readable; add comments if complex
  • Avoid overly broad patterns that cause false positives
  • Include patterns in both English and Spanish when applicable
  • Place the pattern in the correct severity level

Code style

  • Follow PEP 8
  • Use type hints where possible
  • Keep functions focused and well-documented
  • Test your changes against repos with known benign content to check for false positives

Severity levels

Level Score Use for
critical 10 Credential exfiltration, system prompt override, prompt reveal
high 8 Jailbreaks, impersonation, code execution
medium 5 Subtle manipulation, hidden HTML/scripts
low 2 Secrecy indicators, ambiguous patterns

Questions?

Open an issue with your question. All contributions and feedback are welcome.