Policy on the Use of LLMs

We recognize that today LLMs have become widely accepted as tools to support software developers in their day-to-day work and even their free time open-source contributions.

Our policy is it henceforth, that contributors are free to use whatever tools they wish to craft their contributions, as long as a human is responsible for both the initial creation of a contribution (the prompting) as well as reviewing the output of their tool.

The contributor will always be seen as the author of their contribution and are thereby always responsible for the entirety of their contribution. Contributors must be sufficiently confident that the contribution is compliant to all formal requirements and best practices to be able to contribute to this project; and that this contribution is inherently worth bringing into the project. Additionally, the must be able to answer questions about their work during review.

Importantly, this policy forbids agents which take action without human supervision. Similarly, automated review tools which post comments or partake in discussions are also banned.

Transparency Requirements

Contributors must be transparent and label contributions that contain any amount of tool-generated content. This label must be present in the commit message as a message footer and must specify the exact tool name and model version.

Example: Assisted-by Format For Commits

Author: CoolCollab <cool@collab.com> 
Date:   Wed Jul 22 15:53:05 2026 +0200

    Feat: Some Coool Feature
    
  
    Signed-off-by: CoolCollab <cool@collab.com>
    Assisted-by: Mistral Small 4 (mistral-ai/mistral-small-4)

If multiple tools are used, multiple Assisted-by blocks are necessary.

For issues, discussions or any other kind of community post, an alternative suitable label can be applied.

Example: Labeling AI assistance in Communication

In issues, you can add a [AI-assisted] tag into an issue description.

In comments, posts or emails to a maintainer, please specify whether your findings were created with the help of AI, and to which extent.

I used Mistral Small 4 (mistral-ai/mistral-small-4) to help answer this question.

This transparency helps us understand the role of these tools.

This policy includes, but is not limited to, these types of contributions:

  • Code contributions in any form (including CI configuration, build scripts and tests)
  • Design Proposals or Discussions
  • Issues or Security Vulnerability Reports
  • Comments and Feedback on PRs
  • Documentation

Rules for LLM Assisted Reviews

As stated in the beginning of this policy, any LLM or AI agent that automatically reviews contributions or discussions without human intervention is not allowed.

However, contributors and maintainers are allowed to help them review individual parts of contributions. These assisted reviews are only supposed to be used on individual code snippets and annotations in PR reviews and must be labelled clearly as laid out above.

Rules for AI Training

The Nazara Project does not allow any part of our codebase, documentation or community interaction to be scraped and used for training or building AI systems.

We do not cooperate with any individual, group or organisation that does so.

Prohibition of "AI"-generated Artwork, Music or Other Media

Due to the prevalence of stolen artwork in the training sets of generative AI, we do not accept any contribution that includes AI generated imagery, music or other kinds of media.

This prohibition extends past classic code or documentation contributions into our community spaces such as Issues, Discussions and any avenue of communication that already exists or may in the future be introduced.

No exceptions.

Restrictions

We do not allow AI agents to be used for contributions to issues labelled as good-first-issue as these are non-critical and are meant to be learning issues for new contributors and programming beginners.

Of course this does not cover new contributors using LLMs to understand a concept before wirting their contribution by hand.

Enforcements

We reserve the right to reject or revert any contribution - in any form - which is found to violate this policy.

On repeated offenses we also reserve the right to ban any individual, group or organisation from contributing to Nazara or any project of the Nazara Project organisation.