AI-Generated pull requests for new features.

[mattwoberts]

Hi all.

Just wanted to start a conversation here about some rules we're probably going to enforce for new pull requests.

We've noticed that the volume of pull requests has increased slightly, which is of course a good thing. It's not really much of a surprise to see that a lot of these now are driven from AI - usually claude. We're not a "no AI" camp here, if reviewed carefully and the code changes are understood, bring value, and are worth the maintainability cost, then AI can bring in lots of changes much faster.

However what we're started to see are features that are being developed and PR-pushed, that aren't part of what we might call a roadmap. Often there's no related suggestion on our feedback board over at https://feedback.fider.io - so there's no sense that this is something that would value other users. There's also no related issue with discussion.

The burden on me (and it's currently just me) to review and take responsibility for these changes to come into the codebase is quickly becoming too much. It's all nicely summarized in this post here https://www.theregister.com/software/2026/02/03/github-ponders-kill-switch-for-pull-requests-to-stop-ai-slop/4334869 - the burden has shifted from the coder to the reviewer.

So - not wanting to be all down on pull requests, because they're generally a good thing, I am proposing the following going forwards.

• New features need to be related to a suggestion in our feedback fider board. That suggestion needs to have sufficient discussion and votes to warrant its development. If you're planning to develop a solution then create an issue in github and reference the suggestion in fider, and wait for feedback confirming that we're happy to see this happen.

If a pull request appears for a new feature and it's not something that has been discussed and generally accepted onto the roadmap for Fider, then we'll close it.

Hope that makes sense, happy to hear any opinions on this