It really is underdeveloped.
The duplication alone is awkward.
And those > menus seem to be a cover for poor planning.
The word New appears 13 times.
That's alot.
What’s wrong with this community ?
- Thread starter Semper
- Start date
The duplication alone is awkward.
And those > menus seem to be a cover for poor planning.
The word New appears 13 times.
That's alot.
..and yes there are other design issues.
We simpified i, the only part i haven't cracked is, why do we need the thread title and post button there... it's awkward, looks like a visual glitch, and wastes horizontal space. We'll fix it in 2.4.
Hope team Xenforo sees this and copies our design
How do you define "AI slop" or "vibe coded" ?
In my mind 'AI slop' is the potential result of 'vibe Coding', the opposite of what was suggested. I wish there was some way of defining or perhaps more importantly recognising vibe coded programming because we are still at the point where it's not simply not good enough and should be subject to expert oversight and testing.
I'm sure that will change in time but right now I would want to know if AI has been used simply as a developer tool or for the entire programming process.
Looks like version 1.5.24...
What might help is if you get the core style sheet from @Russ site pixelexit it gives that affect too.
What's the difference between the two?
I hang out in a chatroom full of senior professional software developers who get paid large sums of money to build complex enterprise-level applications. None of them would call themselves vibe coders. Most would probably describe what they do as "agentic coding".
It's rare that any of them actually do any coding by hand anymore - they're still producing high quality code that is absolutely enterprise-ready, but they almost never type the code itself, it's all generated by an LLM under their instruction or following their specifications. Most of them are still doing code reviews manually (actually reading the code) - but even that is starting to be automated in many ways - we regularly have discussions about code reviews and proving code quality and approaches are evolving rapidly.
To an untrained eye, what these professional developers do is indistinguishable from "vibe coding". It takes someone who understands how software development actually works and has a deep understanding of agentic coding to recognise the difference.
FWIW, I do absolutely consider there to be a difference between vibe coding and serious software development or agentic coding - but my point is that most people can't tell the difference because they don't understand what it is that modern software developers actually do.
So I'm genuinely curious about how people think they can tell the difference.
Obviously the "quality" of the output is a huge indicator - but is there really that much difference between a bad or inexperienced coder releasing something hand-written and vibe coder releasing something written by an AI agent? Indeed, given how good the AI agents are now, I'd actually suggest that the code they're generating today is better quality than what all but the most experienced developers can do by hand. I know I've recently had my AI agents find a lot of bugs in code I've been running in production for over a decade now!
Ironically, if you follow the current best practices for agentic coding - noting that those best practices have changed significantly over the past 6 months due to the rapid evolution of the coding focused LLMs like Claude - you'll note that their recommended approach to development processes is to actually step up a layer and let the LLMs make more of the decisions about how they're going to achieve your stated goals (ie don't try and micro-manage them as much), which, ironically, makes the best practice look far more like "vibe coding" than it ever has before
No they aren’t.
We triaged nearly 1,000 bug reports over the course of around a week. Some of which were closed. That’s the first and only time that will need to happen at such volume.
I considered turning off what’s new or suppressing the updates but as it was a one-time thing, it didn’t seem worth it.
Honestly, no worries on my end. The What’s New feed is really handy, and I don’t mind the update threads at all. I like to read through it daily to see what’s happening, so a busy stretch like that doesn’t bother me.
Thanks for working through all those reports!
Thanks for working through all those reports!
What's the difference between the two?
I hang out in a chatroom full of senior professional software developers who get paid large sums of money to build complex enterprise-level applications. None of them would call themselves vibe coders. Most would probably describe what they do as "agentic coding".
It's rare that any of them actually do any coding by hand anymore - they're still producing high quality code that is absolutely enterprise-ready, but they almost never type the code itself, it's all generated by an LLM under their instruction or following their specifications. Most of them are still doing code reviews manually (actually reading the code) - but even that is starting to be automated in many ways - we regularly have discussions about code reviews and proving code quality and approaches are evolving rapidly.
To an untrained eye, what these professional developers do is indistinguishable from "vibe coding". It takes someone who understands how software development actually works and has a deep understanding of agentic coding to recognise the difference.
FWIW, I do absolutely consider there to be a difference between vibe coding and serious software development or agentic coding - but my point is that most people can't tell the difference because they don't understand what it is that modern software developers actually do.
So I'm genuinely curious about how people think they can tell the difference.
Obviously the "quality" of the output is a huge indicator - but is there really that much difference between a bad or inexperienced coder releasing something hand-written and vibe coder releasing something written by an AI agent? Indeed, given how good the AI agents are now, I'd actually suggest that the code they're generating today is better quality than what all but the most experienced developers can do by hand. I know I've recently had my AI agents find a lot of bugs in code I've been running in production for over a decade now!
Ironically, if you follow the current best practices for agentic coding - noting that those best practices have changed significantly over the past 6 months due to the rapid evolution of the coding focused LLMs like Claude - you'll note that their recommended approach to development processes is to actually step up a layer and let the LLMs make more of the decisions about how they're going to achieve your stated goals (ie don't try and micro-manage them as much), which, ironically, makes the best practice look far more like "vibe coding" than it ever has before
There is a huge difference between someone who has a technical background (not even programming, but is familiar enough to know pitfalls to avoid) and someone who is just sending prompts to a LLM. I do not find anything inherently bad with vibe coding, however if you are releasing something that is vibe coded it should 100% be clear that it was, and that you may not be the most technical when it comes to supporting the product.
As far as knowing who vibe codes and who doesn't, I go by the following:
- If they have no history I am going to assume they're vibe coding.
- If they have no history of technical or developer discussions (outside of being a customer asking for support) I am assuming they vibe coded.
- If they have never released an add-on prior to ~2025 I am assuming it was vibe coded.
- If they previously only released simple add-ons/template edits, and are now releasing complex add-ons I am assuming they vibe-coded.
- If they are releasing multiple add-ons in a short period of time that would be an unreasonable pace for a skilled developer, I am assuming they vibe coded.
- If their support responses sound like copy/paste LLM responses, I am assuming they vibe coded.
I get that because i've built some serious stuff with AI by now, but it's only made possible because i know how to program still ( for how long? ), do intensive systems thinking, and sweat all the details.
My problem with things that look AI coded is that i don't know anything about the competency behind the person using it.
Usually they don't have the talent and i'm about to waste time being someone's unpaid alpha tester.
I will use AI generated software if i know the developer has serious chops though from prior activities because you know that, at the very least, most of it will work.
Anything else? In the words of 'sweet brown'
There's been a flood of forum add-ons released recently that need multiple revisions which suggests to me they have been vibe coded with little or no oversight. Granted I've not examined the code and these days I would probably struggle to do so but my perception is that these are not add-ons being hand-coded by novice coders.
I also have some familiarity with the drive to introduce AI coded routines within the actuarial profession which has a history of adapting itself to technological advancement but right now it's just not happening and to the best of my knowledge very little has been produced that can be trusted at a production level.
I do take generally agree with the points you've made though and I'll also admit to a bias, distrusting anything produced by AI but that's a story for another day.
From my professional perspective:
The focus should be on code quality, security, maintainability, and reliability, not on whether AI was involved in producing the code. In fact, the thoughtful use of AI in software development can be a strong indicator of a commitment to quality, efficiency, and continuous improvement.
Ultimately, what matters is not whether AI was used, but how effectively its output is reviewed, tested, validated, and maintained.
- No software should be installed without first being reviewed with the assistance of an LLM, or without a reliable assurance that such a review has already taken place.
- AI has become an integral part of modern software development, and the vast majority of professional developers now use AI tools to write, review, debug, and optimize code.
The focus should be on code quality, security, maintainability, and reliability, not on whether AI was involved in producing the code. In fact, the thoughtful use of AI in software development can be a strong indicator of a commitment to quality, efficiency, and continuous improvement.
Ultimately, what matters is not whether AI was used, but how effectively its output is reviewed, tested, validated, and maintained.
We have now arrived a long shot away from the original forum topic in one that has already a long thread since May this year, when people recognized the arrival of loads of vibe coded add ons of "debatable" quality:
Most of what is written here has been written there as well already. The one thing that has changed is that within less than six months it seems that basically most devs (including the very reputable ones) are using AI for coding to a massive amount regularly. This has been expected, but maybe not as fast.
Most of what is written here has been written there as well already. The one thing that has changed is that within less than six months it seems that basically most devs (including the very reputable ones) are using AI for coding to a massive amount regularly. This has been expected, but maybe not as fast.
I commented something similar elsewhere on the forums in early-mid-2026: If AI was/was-not used to assist-with or solely-used to develop an addon/resource, a tick-box or item selection box needs to be available to depict AI usage - and making it a required item.
Sorry @Painbaker - used your addon as an example for this, it was the first item on the list.
For example, if the resource had no AI utilized whatsoever, this could be the result:
If the resource was coded without the use of AI, but had AI assisting with the process, this could be the result:
If the resource was created with AI and AI alone, without a human doing any manual coding, this could be the item displayed:
Why should something like this exist?
If the addon author coded something by AI only:
1. There's a chance that future updates for the code will be slim to never.
2. There's an elevated risk for the addon to have issues when installing, removing, upgrading.
3. There's a very real chance that the addon developer does not know PHP very well, and could be insecure or unopimized.
4. Addon authors performing going this route might be in the mindset of quantity and profit over product and quality.
If the addon author coded something and then had AI assist with the code, or rounding off the code:
1. The addon author knows what and how the addon should function.
2. The addon should be able to receive updates and bug fixes in a timely manner.
3. AI might amend code, and introduce spurious issues/exploit vectors if not caught by the addon author.
If the addon was coded by hand, without AI usage:
1. Addon author knows how to code in PHP (and other related languages/syntaxes).
2. Knows the framework of the addon and its functions exceptionally well.
3. Addon updates might be slower than AI assisted addons.
But hey, just my 2c on the matter.
After watching a news item on robotics in surgery, I’m sure we will feel that similar questions should be asked:
- Will my surgery be carried out solely by a robot?
- Will it be a human directed robot?
- Will it be a 100% human surgeon
I'm so hesitant on robot-only surgery, let alone human-controlled/directed robot surgery... that is a massive rabbit hole in its own regard. I can see the use-case for it, but damn... I'm so conflicted with that. Robot has a motor failure? You could be injured. Robot loses power for some reason? You might be injured. Surgeon doing the operation loses internet connection? Robot might do something awkward or freeze-in-point. It's quite the conundrum.
AI is nosy technology that people are trying to make us embrace it.
If i ever decide to use AI it will be for something positive such as registration.
You only have to go on any meta platforms to get a survey about some ad that's come on your feed asking if you have bought anything from it.
It's like somebody has played the same prank on you and you're sick of it.
If i ever decide to use AI it will be for something positive such as registration.
You only have to go on any meta platforms to get a survey about some ad that's come on your feed asking if you have bought anything from it.
It's like somebody has played the same prank on you and you're sick of it.
FWIW, Amazon KDP (their self-publishing platform) has a required prompt for two check boxes (wording paraphrased).... (A) Was this created with AI? (B) Was this created with assistance from AI? It doesn't affect the listing or have a negative impact but the disclosure is required.
If similar was required here for the RM submissions, with no judgement applied, just the disclosure of whether it was created fully with AI or created with assistance with AI, I think that would help with some of the concerns.
I'm actually starting to enjoy seeing the XenForo Bug Bot appear on the What's New page. Plus, seeing bugs changed to fixed is awesome. For the longest time, there was no progress, but this is a step in the right direction. This shows progress is being made, and we're that much closer to a new patch or a new release.
Well said.
It seems AI coding has proven itself, and it seems the tipping point was pretty recent.
Similar threads
