3.2 KiB
AI Usage Policy
Summary
Usage of AI tools is authorized in this project.
However, you are accountable for all produced code and must ensure you properly validated it before committing !
Scope
This policy covers all contributions to this repository: source code, tests, documentation, configuration, commit messages, issues, and pull requests.
What is allowed
- Generating, refactoring, or explaining code with any AI tool.
- Writing tests and PR descriptions.
- Make extra sure you review those in depth !
- Opening pull requests, as long as a human reviews the result before merge.
- Debugging, code review assistance, and research.
What should be avoided when possible
- Writing every bit of documentation with AI.
- Using AI to write commit messages.
- Keep it as human and clean as possible
What isn't allowed
- Using AI to generate any artistic assets of any kind.
- I'd rather have ugly assets than soulless AI shit.
- It's a shitstorm I want to stay far far far away from.
- Using AI to open bullshit issues that are irrelevant and waste time.
The Core Rule: Human Validation
AI output should be treated as a draft.
Before a change is merged, the submitting human must:
- Read it. Every changed line, not a skim.
- Understand it. You can explain what it does and why, without the AI.
- Verify it. Build and test it, make sure you don't break it, and update what needs to be.
- Own it. You would defend it in review exactly as if you had typed it yourself.
If you cannot do all four, the change is not ready.
Reviewers
- Treat AI-assisted changes to the same standard as any other change.
- An AI review alone does not count as an approval.
Licensing & accountability
The human submitter is the author.
- You will be fully responsible for any security or licensing issue.
- If required, you must be able to provide the prompts that led to your code.
- We will aggressively discard contributions when licensing issues arise and you can't do that.
Restrictions
- Secrets and sensitive data: never paste credentials, keys, personal data, or confidential customer data into AI tools that are not approved for that data.
- Licensing: do not accept output that reproduces third-party code with an incompatible license. If a suggestion looks copied, treat it as copied.
Disclosure
Any commit mostly written by an AI must be clearly indicated as such with an appropriate Co-Authored-By section in the commit's message.
Example:
Co-Authored-By: {AI's name} <{AI's email}>
Enforcement
- Changes that violate this policy will be reverted.
- Repeated violations will result in blanket refusal of any further contributions.
Author's note
I hate having to write this kind of moral sermon shit, but the current landscape of software development is evolving fast, and I can't be fucked to deal with bullshit drama around the usage of AI.
If you don't like it, fine, go away and make your own.
If you like it, feel free to burn some tokens around the bonfire with us.
Changes to This Policy
This policy is maintained by Herwin Bozet (NibblePoker).
Feel free to propose changes via a pull request.
Last updated: 2026-11-03