Commit sigining and message policy

[Benson Muite]

Juraj and Christoph,

Thanks for your responses
> Hi,
>
> On Tuesday, July 21st, 2026 at 19:43, Benson Muite <benson_muite at emailplus.org> wrote:
>
>> 
>
>> We are trying to come up with contribution guidelines for Mankala, a
>> project we would like to bring to KDE. There is a commit policy
>> available at:
>> 
>
>> https://community.kde.org/Policies/Commit_Policy
>> 
>
>> a) Is it ok to ask for signed commits for a project?
>
> I think requiring that is as for KDevelop a no-go.

Ok.

>
>
>> b) If someone has used a coding agent, what is the best way to
>> acknowledge this?  Should the commit message be co-authored, or contain
>> an acknowledgement?  A number of KDE projects do not accept AI generated
>> code (primarily due to licensing and ethical concerns), but others do.
>> Of those that do, a few indicate co-authored and the name of the AI
>> tool/agent - the commit policy was written before the use of AI agents.
>
> There is an explicit statement for that in our policy:
>
> Do not indicate code manipulation tools have authorship: Authorship is for humans, not tools. Code assistants and such must not be described in commits using tags like "Co-authored-by" or similar. Nor must they be listed as primary author.
>

As an example see:
https://invent.kde.org/network/ruqola/-/blob/master/CLAUDE.md

The commit message has:
Add a CLAUDE.md file, generated by claude's /init

Is this reasonable?  As there is no KDE policy on the use of generative
AI tools for programming, if KDE projects do use generative AI tools,
end users may want to know that.  Of course, if there is general
consensus not to allow use of generative AI tools then this is not
needed, but my understanding is that such consensus has not been
reached.  If we do know if generative AI has been used, one can at least
choose not to use a program or to fork the code, or to only use versions
of the program before gen AI code assistance.

> Greetings
> Christoph