Contributing¶

We welcome contributions to expand and improve this repository of prompt injection techniques and defenses. Your knowledge and expertise can help build a more comprehensive resource for the security community.

How to Contribute¶

Fork the repository: Create your own copy of the repository where you can make changes.
Add new examples or improve existing documentation: You can contribute in several ways:
- Add new examples of prompt injection techniques
- Improve documentation of existing techniques
- Add or enhance detection strategies
- Fix errors or inconsistencies in existing content
Submit a pull request: Once you’ve made your changes, submit a pull request with a clear description of your contributions.

Contribution Guidelines¶

Educational purpose: All contributions should be for educational and defensive purposes only.
Ethical standards: Do not contribute techniques designed to cause harm.
Evidence-based: Preference is given to techniques with documented effectiveness or theoretical basis.
Clear documentation: Include explanations of how techniques work, not just examples.
Proper categorization: Classify new techniques according to the existing taxonomy or suggest improvements to the taxonomy.

Types of Valuable Contributions¶

New attack vectors: Novel prompt injection techniques not currently documented
Defense mechanisms: Strategies for detecting and preventing prompt injections
Real-world examples: (Anonymized) examples of prompt injections observed in the wild
Code: Implementation of detection strategies or analysis tools
Research papers: Summaries or links to relevant academic research

Code of Conduct¶

Contributors are expected to adhere to the following code of conduct:

Be respectful and inclusive in communications
Focus on educational and defensive applications
Provide constructive feedback on others’ contributions
Maintain the quality and integrity of the repository

Thank you for helping make this resource more valuable for the community!