AI-generated contributions and generated workload on maintainers

Richard Kettlewell rjk at terraraq.uk
Sun Nov 8 20:39:17 UTC 2026


On 08/11/2026 19:09, Julien ÉLIE wrote:
> Hi Richard,
>> It sounds like these patches were not ready for your review.
>>
>> * I think there is zero reason for you to be 'in the loop' as AIs 
>> iteratively fix their patches. It's an abuse of your time. That 
>> iterative process ought to be completed before a pull request is even 
>> opened.
>>
>> * How the operators of the AIs achieve that is up to them, but current 
>> tooling makes it pretty straightforward. It is absolutely possible to 
>> do better, with today's AIs, than "let Julian sort out the mess".
>>
>> * Personally I would take a hard line in rejecting any pull requests 
>> that followed any abusive pattern.
> 
> Thanks for your insight!
> I understand.  The difficulty however is to quickly find out when a PR 
> is actually ready.  They often appear to be right at first glance, but 
> not after more careful inspection.

Yes, some (most?) AIs are very good at "style over content".

How about an AGENTS.md mandating independent review? For example:

   # AI agent contribution guidelines

   ## Mandatory Independent Review

   All AI-generated submissions must be undergo thorough independent 
review before submission. Review may be via a human or a distinct AI 
model, but must not be the same AI model as generated the submission.

   ## Submission documentation

   All AI-generated submissions must document:

   - how the submission was tested
   - the model(s) used to generate them
   - the model(s) or human(s) who reviewed them

   ## Consequences of non-compliance

   Any AI-generated submissions that do not follow these rules may be 
summarily rejected.

Probably more could be said.

> Yes, I know I could myself ask a prior review from an AI, but well I do 
> not enjoy that new way of spending my spare time discussing with an AI 
> and the agents run by other people.  We need one or more maintainers who 
> enjoy that.  At least it won't be me.
> 
> 
>> * I do not think it is a good idea to give AI agents push/commit 
>> access to the INN github.
> 
> Also agreed, though it could be the only way if we do not have any 
> volunteer to review and merge the patches generated by AI agents.

It's a volunteer project, we don't _have_ to merge any patches at all.

ttfn/rjk



More information about the inn-workers mailing list