GPT-4.1 in Copilot is Driving Me Mad – Where Did the Commonsense Go?!

[zairulanuar]

Select Topic Area

Question

Copilot Feature Area

General

Body

I don't know who needs to hear this at GitHub HQ, but this ain’t it. I’ve been using GitHub Copilot since the early days, and now with GPT-4.1 in Copilot… I’m at my wit’s end.

It’s like babysitting a junior dev with amnesia. Commonsense? Gone. Diligence? Missing in action. Every time I fire up a task, I’m not getting a copilot, I’m getting a barely-functioning autocomplete with vibes.

Here’s what’s breaking me:

I literally have to point it to the related files manually. Like, “Hey buddy, yes, check the controller too—no, not the view, the controller. Yes, again.” Why do I need to spell it out like I’m explaining recursion to a potato?

It refuses to double-check anything. It hallucinates code, forgets structure, skips validations, and worse—acts confidently wrong.

I’m in an agent-based workflow, expecting some sort of context awareness. Nope! I’m the one doing all the connecting, error tracing, and file referencing.

And don’t even get me started on database checks—it won’t trace relationships, it won’t look up schema unless I handhold it like it’s on training wheels.

Agent mode? More like Agent Nope. I spend more time cleaning up its lazy suggestions than writing actual code. It’s gone from “superpower” to “super-annoyance.” I used to fly through problems with Copilot. Now it’s like dragging an AI intern who keeps asking where the bathroom is.

Look, I want to love Copilot. I pay for Copilot. But GPT-4.1 is seriously making it feel like I’ve got an overpriced text expander with delusions of grandeur. If this is the future of AI pair programming, then I’m out here living in a bug-filled time loop.

Bring back reliability. Bring back smart file linking. Bring back commonsense! Or at least give us the option to use a better model.

Signed,
A pissed-off dev who just wanted a little help and got an AI anchor instead.

[varshith999]

I’m working on a college project and using Copilot with GPT-4.1, but honestly, it’s become super frustrating. Earlier it used to help a lot, but now I feel like I have to explain every small thing to it. It doesn’t understand the flow between files, gives wrong code, and misses basic logic. Instead of helping, it’s slowing me down. I really hope GitHub brings back the older version or fixes these issues soon.

[KristinaWhites]

Hi, yes I get it now I don't know if having premium madders when it comes to this but, I was trying everything and us getting basics mistakes and I didnt know how to code not one bit. Now I kinda understand what's going on a bit but I digress. I found the option to use and Agent in particular Claude Sonnet 4.1 in vscode thru Github and my whole world changed I asked it to do something for me and it did back to front I was like wow. Now that doesnt mean we don't spend hours debugging cause we do. Now I'm not sure what your project is but an agent is way more, lets say , global it does it all. the mistakes GPT-4.1 was making were on the html-css- java forget about it. Ask it to fix something it calculates and doesn't do what you asked might break something else. I'm sorry I think the difference is big. Claude whipped out a whole new style made no mistakes in the html or css java was a little back and forth. It rights all the php files to get mySQL going and here is we are stuck a bit, like for a day or so but I have faith Claude and I will get to it and then back will be connected to front and then there is no limits to what I can do with claude. One web app after the next. I just nee to come full circle back to front and its my opinion GPT-4.1 I'd still be trying to get my landing page and css right. I had gave up and was out there learning to do it myself tell I met Claude <3 hearts ...hope that all makes sense maybe it helps.

[anchildress1]

🤣 🤣 🤣 I frequently compare GPT to a 2 year old with hands into everything, chocolate smeared on every wall, marker all over every inch of it's body, with a trail of toilet paper highlighting it's every step along the way. I use more CAPS trying to tame it enough to just fix the stupid test, "NOT THE CODE, THE CODE IS FINE! The TEST is broken. HERE #selection, can you fix it now?" and it alerts me at least 10 times that it's identified the real problem and it's definitely fixed, when we all know assuming the test was even touched to begin with (it probably wasn't), it's still definitely not fixed.

Solving those sorts of coding problems is not it's strong suit in the least. It is, however, really good in Ask mode when you're in those early planning stages of development and want to explore all the possible ways to achieve a goal. Or explain the pros and cons for a potential development plan. That part it can really excel at, so I usually forgive it for the 10 minutes I just wasted and the corresponding momentary yet substantial increase in blood pressure.

[XXOriginalXX]

Honestly, I can’t tell if chatgpt and Copilot got dumber or if my projects leveled up so hard they left them behind either way, debugging with them lately feels like babysitting two interns with amnesia

[turbohacks]

please repyll me

[nexantdashley]

Yep, I don't know what they did since introducing the paid stuff, but what before was awesome, is now awful. Not just mediocre. It's infuriatingly bad. Like so bad I'm about to quit using it altogether, which is very sad, because I had really come to love it and would have paid to keep it. Is that what I'm supposed to do, start doing what I did before but as premium requests now? If so, that's fine, but that wasn't clear to me in the communications that went out.

[Tremendous69]

I agree. It feels like I have to correct him evey time it makes a change. It is just lazy and poor compared to claude. No thanks, I’m too frustrated

[ashleyleslie1]

I feel like GPT 4.1 in github copilot is the same as the auto complete stuff we had 2 years ago... literally can not complete ANY features or fix problems with GPT 4.1 and I have to keep doing things manually. Still waiting for my premium requests to reset so I can use at least Claude Sonnet.

[addshore]

This all also summarizes my experience sticking on 4.1 for the past days.
Sad my Claude Sonnet 4 requests are all used up, as it really seems to do a better job

[linonetwo]

Gpt4.1 keeps asking me if it should continue, I SAY CONTINUE UNTIL PERFECT!!!!!!!

Claude won't do this, claude is always working hard till the end.

[hariii15]

[linonetwo]

When the world only have gpt, we won't complain. But now we have claude for competition, then we fell its bad.

[jantoney]

@hariii15 - I disagree. I can do the same prompt with 4.1 and Sonnet 4 and sonnet will power through, plan well, code, test and clean up very well.

4.1 is like a baby. It might plan, but then it just dumps a bunch of <script> code into the middle of a CSS style block or put a function into the middle of another function.... both literally happened to me today.

I use Sonnett 4 the most now, flipping back if I'm just trying to do basic things and save on premium requests.... But it doesn't last long because 4.1 you need to almost go find the files it needs to work on. and then if it asks you a question and you reply to it. It completely forgets the context and acts like you've asked it a brand new question.

I'm now very good at prompting I think. I can 1 shot prompt a large complex feature addition to an app and have Sonnett get it right almost every time. So it's 100% not me. It's defs 4.1.

Maybe 3-4 months ago 4.1 was amazing. Something has changed since introducing paid requests though. I feel like they don't pass as much context over anymore just to force it to use less tokens - ergo keeping costs down and allowing them to do unlimited 4.1 requests - but paid Sonnett

[ashleyleslie1]

p{margin-top:0;margin-bottom:0} I am explicit, it works if I use sonnet 3.5 or even other version of gpt. The problem I am describing is explicitly the use of unlimited gpt 4.1 through GitHub copilot. Sent with DaumMail App

…

--------- Original Message --------- From:hari ***@***.***> To:community/community ***@***.***> Cc:Ashley Samantha Leslie ***@***.***>,Comment ***@***.***> Date:Tue Jul 08 18:12:40 GMT+02:00 2025 Subject:Re: [community/community] GPT-4.1 in Copilot is Driving Me Mad – Where Did the Commonsense Go?! (Discussion #163630) Not really though, i think the use of ai is making the developera too lazy, your problem is the answer for most of the guys out there, you are not being explicit, clear and precise, Not to offend or anything but, ai is an powerful tool, we just need to make a proper use of it, considering 4.1 is Opensourced too Try going pro or something. —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you commented.Message ID: ***@***.***> [ { ***@***.***": "http://schema.org", ***@***.***": "EmailMessage", "potentialAction": { ***@***.***": "ViewAction", "target": "#163630 (comment)", "url": "#163630 (comment)", "name": "View Discussion" }, "description": "View this Discussion on GitHub", "publisher": { ***@***.***": "Organization", "name": "GitHub", "url": "https://github.com" } } ]

[hariii15]

Yeah that happens a lot, the models are not good at styling, when the files are large most of the models fail at chamging minute cha ges in the file, in that cases ill do them my self though, lol i think about vibe coders every time in times like that

[neptunusRomulus]

All GPT models in Copilot really sucks. Claude really works well but it got a limit in a day, after that you had to work (fight) with the GPT models... It can't even change a position of a simple button in a jsx file.

[Metanome]

Copilot officially sucks now after the rate limiting, it feels like we're moving backward instead of forward.

Either way, I was recently advised to use Copilot with RooCode if you have been rate limited, have you tried it? Also, there's Gemini Assist now; have you tried that?

[jantoney]

Gemini is ok - the CLI interface sucks though and the file changes interface on CoPilot is really unmatched.
Since vsCode is opening up all the interfaces to allow others to use them - I feel like that will fix these issues moving forward though.

[Metanome]

I'm not sure if you're referring to both Gemini Code Assist, which is a VSCode plugin, and Gemini CLI, which I haven't tried yet. However, Gemini Code Assist looks promising, especially given its generous daily token usage. Aside from that, you can use any AI API with RooCode, and it works wonders, much better than what GPT-4.1 can do on its own within Copilot.

[alistair-riskmemo]

I've given up with agent mode entirely. It always produces basic code errors, like missing curly brackets, and I'm never sure if it's going to randomly remove an import, for example, for no apparent reason.

The chat still helps, sometimes, but I'm very close to cancelling my Copilot subscription.

[jantoney]

Use Sonnett 4 instead of 4.1 - that will get you back on track.

Also, try go to GPT online and say - I'm trying to write a prompt for COPilot AGENT mode to make a change. Here is what I have.
Dump in your prompts, then let it restructure the request and use that. It's a great way to learn how to structure the prompts - you'll quickly pickup the good way and will be able to write them quickly with few mistakes.

[apete]

I've used both copilot and cursor for a while, and my general impression is that they're getting worse. I think it's some form of agent hybris. I get much better help just using ChatGPT detached from any IDE.

[kashima19960]

在copilot chat中使用GPT4.1，你可能会遭遇以下要素

1. 精神胜利：当你以agent mode要求GPT4.1完成一个文件修改数量很多的任务并且要求他完成任务后编写修改报告时(比如10个.c和.h文件以上)，他会读取所有你要求的文件，然后直接跟你说：“我已经完成了所有的任务”，但打开git工作区一看，没有任何修改记录，呵呵，我不知道为什么GPT4.1能如此自信，在精神上胜利就行了是吧？😅
2. 正确的废话：用GPT4.1解释代码更是一点用处都没有，十分喜欢说一些“非常正确的废话”，看似说了一大堆，但是实际上一点用处都没有
3. 比牛还倔强：当你给出一个稍微复杂一点的任务时，他大概率都会回答”我已经了解你的全部需求了，请稍等，我马上就会开始着手任务“，然后我一看，输出终止了，不是，哥们，你都明白了些啥啊？明白了就一直输出内容啊，难不成要我一直告诉你”谢谢您，尊敬的GPT4.1大人，请您尽快开始任务吧“
4. 听不懂人话：当你给出一个简单但是工作量比较大的任务时，GPT4.1会一直询问你”如果你还有其他的xxx需要继续完成，请告诉我“，即使我明确指出要完成全部的任务，不能中途一直暂停询问，GPT4.1依然会一直暂停询问，跟一个听不懂人话的婴儿或者精神病院的弱智一样
5. .....

english version:
When using GPT-4.1 in Copilot Chat, you might encounter the following phenomena:

1. The Mental Victory: When you put it in agent mode and ask it to handle a task involving numerous file modifications (say, over 10 .c and .h files), and then request a summary report, it will read all the files and confidently declare, "I have completed all the tasks." However, a quick look at your Git workspace shows that nothing has been modified. Heh, I have no idea how it can be so self-assured. I guess a mental victory is all that matters, right? 😅

2. Correct but Useless Platitudes: Using GPT-4.1 to explain code is utterly pointless. It has a habit of spouting "perfectly correct nonsense." It seems to say a lot, but in reality, none of it is the least bit helpful.

3. Stubborn as a Mule: If you give it a moderately complex task, it will almost certainly reply, "I have fully understood all of your requirements. Please wait a moment while I begin working on the task." And then you look, and the output has simply stopped. Seriously, what exactly did you "fully understand"? If you understood, you should be outputting the result. Am I supposed to constantly plead, "Thank you, your majesty GPT-4.1, would you please be so kind as to begin the task now"?

4. Inability to Follow Instructions: When you assign a simple but large-volume task, GPT-4.1 will incessantly interrupt to ask, "Please let me know if you have other [xxx] that need to be completed." This happens even when I've explicitly instructed it to finish the entire task without pausing for confirmation. It still keeps stopping to ask, like an infant who can't understand plain language or an idiot from a mental asylum.

5. .....

[xkmato]

All TypeScript errors in [page.tsx] have been resolved. The only remaining issue is a minor unused import ([useState]), which does not affect runtime or build. If you want, I can remove this unused import for you. Let me know if you need further cleanup or have other requests!

That is the response I got for asking it to clear the ts error it had introduced. Now bugs have attitude too. It is not just the PMs anymore.

[userofit123]

Has there been any update from anyone working on this? Are there any planned fixes or updates to address anything in this thread?

[RyelBanfield]

For anyone still having issues, please try breaking your tasks down into small instructions. I think a lot of issues come from trying to get 4.1 to do absolutely everything without any input. My experience has improved so much over the past month that I don't even think about using Sonnet 4 that much.

[phenry9999]

Thank you for trying. And I'm sure you've had great success doing this. But when I try this, it still gives me HOURS of debugging horrible experiences wasting my time with excessive code and useless delusions. As an example, I asked it how to center a checkbox column in a telerik grid. Of course I gave it a bit more info, including the page. It's pretty simple, self contained, but still resulted in TOO long wasted on non-existing properties, non-existing methods, properties and excessive code.
Thank you for trying, but I've been told this over simplification as a panacea for all AI prompts. But it's just not the case.
PS Yes, when I give it MORE context, more explanation, more descriptions, it still continues to waste my time.

[userofit123]

This has been suggested by others as well, and I have tried it but the results are the same overall. Even the simplest of tasks sometimes it just cannot do, for example, if I ask it to add a piece of code, lets say even just CSS, and to add it to a CSS file, instead of adding it to any available space in between other CSS, it will add it in the middle of others, breaking the file. Instead of adding it after a closing bracket } it adds it before.

I have had some more "success" with using the ask mode, it actually gives me better answers that I copy and past into the editor, but, then the question is if this is the only mode that actually works decently, why would I pay for it when I can just use ChatGPT or any other chat for free and copy the code from there? I wish someone from the team would address these matters.

[jfullstackdev]

it's as stupid or as brilliant as the one using it, but for me, it's really helpful, you don't ask what AI can do for you, rather you ask AI on what to do,

hesitance to use it originates from the threat it brings but despite all of improvements, even the promised AI Agents will never be a total replacement for Devs, it will encounter edge cases that only humans can solve, it's still limited intelligence but helpful tool

however, there are really distinctions between GPT-4.1 and say, Claude 3.7, they both have their advantages and disadvantages like Claude 3.7 is too aggressive and you will be hitting the rate limit sooner while GPT-4.1, you need to force it to scan the codebase to know more about the context because it tends to be lazy

but given the proper use, they are as powerful, you leverage from a coder to focusing more on the design and business logic of your codebase

I also wrote an article for those interested

https://github.com/jfullstackdev/ai-agents-far-beyond

[samswartz1972]

Sorry for me saying this but it's total fuckin bullshit to pay for such a garbage Ai Agent that is totally garbage and useless where you ask it to crrect 1 error and it makes 246 more and never fixes the error then when fixing errors it like to put fake codes in when asked not to and you even tell it's a restriction and still turns around and does it. I try swtching over to GPT 4.0 and keeps saying I exceeded such garbage I think Github Copilot is going to lose alot of pro users cause in all reality it's not worth the headache of dealing with Ai Agent that messes up more than it can produce .They be better of switching out GPT 4.1 for claude 3.5 or 3.7

[FatihEdu]

Just noticed this. Even gpt-4o (the other free model) is able to read files, but gpt 4.1 just won't. I tell it to go read the file and it tells me to paste the contents. It wasn't always like this iirc.

[hobsRKM]

GPT 4.1 API on open ai works so good. Looks like Git hub copilot 4.1 seems to underperform, either a marketing strategy to lure more users to burn premium requests on copilot. Sad.

[RyelBanfield]

How exactly are you guys using the agent? It's baffling that so many are having such an experience yet some (like me and others) are able to move fast, build features, fix bugs, etc. I work with TypeScript. Maybe it's other languages where issues arise? Maybe some extensions are causing issues? It's so weird.

[huytega]

I understand your frustration — switching to GPT-4.1 in Copilot has been a rough experience for quite a few devs, especially around context awareness and reliability.
Here are a few practical things you can try to make Copilot behave closer to the “pair programmer” you expect:

1. Force better file context

Open all relevant files before prompting Copilot. It only reads files currently open (and a limited number of tokens), so having the controller/model/schema tabs open gives it more context.

If you’re in VS Code, you can use the "Include Additional Context" command (Ctrl+Shift+P → Copilot: Include Additional Context) to manually feed in extra files.

2. Reduce hallucinations & wrong structure

Break tasks into smaller, specific requests instead of broad “write me X” prompts.

Use inline comments like:

// TODO: Add validation for null values before saving user data

Copilot tends to follow explicit in-file notes better than abstract instructions.

3. Use model switching

Copilot Labs / Settings → switch back to GPT-4 Turbo or even 3.5 for certain tasks. Some devs find these more predictable for boilerplate and schema-aware work.

Keep GPT-4.1 for creative/problem-solving tasks, but not for “plumbing” code.

4. For agent-based workflows

Copilot currently doesn’t have true repo-wide “agent memory.” Until that changes, you can simulate it by:

Creating a scratchpad.md in the root of your repo with relevant notes, schema, and relationships.

Keeping it open while coding — Copilot will often read from it if in view.

5. Give feedback to train it

Every time it goes off-track, hit the thumbs-down and write exactly what went wrong. GitHub’s Copilot team does use this data to adjust model behavior.

[hobsRKM]

its not about instructions, github copilot GPT seems a blunder at coding. Im pretty sure lot of them have same issues as you can see from threads here. OPEN AI API GPT 4.1 has no issues and i have tested it so issue is from github.

I just asked a small peiece of funtion refactor, instead of refactoring at the exact same place despite sending as context, it just copied the entier code out of syntax and broke the entier code. Tried the same on open ai GPT 4.1 API worked like charm.

[hobsRKM]

Here is an open bug
microsoft/vscode#258329

[dckimGUY]

Assalamu Alaikum, Greetings

I have limited interaction with the AI. I only use the independent one as an interactive encyclopedia, which is beneficial. Everything else I do manually, mostly because I am not sure of what outcome I prefer. Neither the finished embodiment.

Special Baseline File Entries

There are special files that can be used to effect a baseline for the 'co-pilot' system.

From what I have read, and information that I have gathered, the GitHub 'Co-pilot' program reads those special 'baseline files' from these specific locations:

Per repository at this location:
.github/copilot-instructions.md

Per directory, and or custom instruction set:
.github/instructions/.instructions.md

https://docs.github.com/en/copilot/how-tos/configure-custom-instructions/add-repository-instructions

I assume that everyone is taking complete advantage of those potential 'baseline input' methods.

Human VS. Machine: YAML?

What is remaining is the human vs. machine aspect. The good news is that we can control our inputs, and treat them a lot more like a sequence of command line inputs.

I would like to strongly suggest YAML as an input language because:

• the format is unambiguous
• there is no 'conversational flavour' in the input
• we will still enjoy a conversational output at the prompt

YAML: A human-readable, data serialization language.

Guppie Fishbowl Analogy

I will now proceed to deliver a basic 'guppie-fishbowl' or 'Hoch-Deutsche/Platt-Deutsche' analogy:

Germany is a Country in Central Europe. It is made up of various regions, and the regions are made up of municipalities, which are in turn made up of towns.
It is possible that within each town there might be a specific way of expressing something that is not entirely out of the ordinary.

There may be many such differences. These language differences, while not removing them from being considered a 'german-language', potentially define a form of 'Plattdeutsch' (also known as 'Low-German').

Across that country there could potentially be any number of different regions, with language differences defining regional variations. These regional language groups are each 'Plattdeutsch.

Hochdeutsch: The language of High-German

Hochdeutsch is something too. This is where it gets interesting: Hochdeutsch is essentially the intersection of all 'Platt-Deutsche' variations.

The German upper classes were aware of this, and therefore became adept at speaking Hoch Deutsche: The unifying language understood by all.

So, that's where the 'fishbowl' comes back in. The little fish look so colourful at first, and then eventually, after a few generations, they are all kinda the same colour.

Not entirely different from the German Language metaphor from before.

So, to re-hash the suggestion:

1. YAML prompt entry for inputs that you are able to express exactly, where you are absolutely sure about the structure or work that is desired.
2. Conversational prompt entry for exploratory, encyclopedia like inquiries. (like how I do)

I would wonder whether the language that we must use to interact with the AI prompt can be standardized and formalized at some point.

That may be the subject of a completely separate discussion question.

Can anyone give feedback about using some degree of YAML like syntax as input? Is that good or bad? Or doesn't matter?

[guillaume-mueller]

It constantly interprets what I say in the dumbest possible way.

It is not a copilot if you can't finish your work on time because you had to take care of it.

[hobsRKM]

This is not basic model,to give so much concise instructions just for a simple task.GPT 4.1 is the dumbest model on github. Even a small task just 3 lines of code refactor is blunder.

GPT 4.1 from open ai outperforms github hosted 4.1. Im pretty sure that they want people to use premium requests and burn more for money. (If this is not the case, then they better fix the hosted model)

[DawidKrysiak]

Over the last few days, CoPilot is driving me MAD. It's not LLM issue, it's CoPilot issue -the same problems regardless of the model.
Instead of actually helping, its displaying snippets with an attitude 'here, go and fix it' rather than actually helping.
Over the previous week, it went from an overly attached girlfriend attitude of generating AI slop of epic proportions to delivering NOTHING.

It's failing across the board - simple Python issues, Terraform templates... solutions are not just 'not good enough' - they are actually breaking things. What is going on?

[M-S-B-A]

Exactly right

[Absiinator]

For me it's now gpt-5 mini that i find is the overall best model for now.
Sometimes, when the context seems to be approching full token usages, it switches languages or navigates through both french (which is my native language) and english (likely to be what it was mostly trained on)
4.1 is just so bad for now, that I rather switch to the previewed 5 that is FAR better in all cases.
I think they should just switch to using it by default rather than 4.1.
And as far as I know it is overall less costly to deploy too.
4.1 was making doubt the usefulness of copilot, 5 just makes me use it more, with better results.
If you have the option, you should try it.

[DawidKrysiak]

yes, but the issue is not with the models - as mentioned earlier - regardless of the model selected, 2 weeks ago Copilot was semi-useful - a lot of slop, but it was interacting with the terminal, proposing changes, insisting on creating test scripts and running them. Then it simply turned 180, and now it's suggesting clearly broken/invalid code. On repeat.
in the end, Copilot burned all my credit and now switched to 4.1 and is pretty much baseline useless.

[dckimGUY]

On the up side:

As long as the error is very obvious, that's better than if it were subtle/profound.

[DawidKrysiak]

haha... yes, but why would I pay for the product that 'sometimes works' but most times causes MORE WORK AND STRESS?
I'm so glad I shorted the positions of all the AI-related stocks - that bubble will burst in an epic display :)

As for this issue - I will wait for the counter to reset next month - there is no way I will pay $50 just to see if it's any better if I throw money at it. Oh well.. it was a good run, it did help me with a few projects, but nothing lasts forever, I suppose.

[RyelBanfield]

"Oh well.. it was a good run, it did help me with a few projects, but nothing lasts forever, I suppose."

Going back to coding everything yourself? That would be interesting. I wonder how that's gonna go a year from now.

[DawidKrysiak]

for me it will mean a bit of a slowdown for now, but not too much. 99% of my work revolves around AWS APIs, so GPT and other LLMs were beneficial in cutting boto3 research time :).

[github-actions[bot]]

💬 Your Product Feedback Has Been Submitted 🎉

Thank you for taking the time to share your insights with us! Your feedback is invaluable as we build a better GitHub experience for all our users.

Here's what you can expect moving forward ⏩

• Your input will be carefully reviewed and cataloged by members of our product teams. 
  • Due to the high volume of submissions, we may not always be able to provide individual responses.
  • Rest assured, your feedback will help chart our course for product improvements.
• Other users may engage with your post, sharing their own perspectives or experiences. 
• GitHub staff may reach out for further clarification or insight. 
  • We may 'Answer' your discussion if there is a current solution, workaround, or roadmap/changelog post related to the feedback.

Where to look to see what's shipping 👀

• Read the Changelog for real-time updates on the latest GitHub features, enhancements, and calls for feedback.
• Explore our Product Roadmap, which details upcoming major releases and initiatives.

What you can do in the meantime 💻

• Upvote and comment on other user feedback Discussions that resonate with you.
• Add more information at any point! Useful details include: use cases, relevant labels, desired outcomes, and any accompanying screenshots.

As a member of the GitHub community, your participation is essential. While we can't promise that every suggestion will be implemented, we want to emphasize that your feedback is instrumental in guiding our decisions and priorities.

Thank you once again for your contribution to making GitHub even better! We're grateful for your ongoing support and collaboration in shaping the future of our platform. ⭐