@spion People are having a serious conversation about the tab-closing strategy. I love this app so much 😂
English
Alexander Furman
315 posts
Alexander Furman
@fav83com
Software engineer
Warsaw, Poland Katılım Haziran 2024
257 Takip Edilen76 Takipçiler
You don't need your tabs. You can close them all. You'll find the important ones later.
Dmitrii Kovanikov@ChShersh
Every week, I forcefully close all my tabs. No matter whether useful or not. This is how I avoid tab hoarding.
English
@arvidkahl Quite interesting that I recently made a skill that does the opposite. It goes through the whole test suite and looks for useless tests or tests that can be consolidated.
English
Okay this might be totally obvious but you can just tell Claude Code to double your test coverage and it will actually do it.
Will need a lot of review, but if you’re just getting started with adding tests to your code, have the AI take a shot.
Highest impact overnight task.
English
This totally resonates with me. I can’t bring myself to use coderabbit or greptile consistently, even though I like how they handle reviews. The whole branch-push-PR workflow is slow and requires too many clicks. At the same time, it’s so easy to write a prompt or create a command or skill to get a decent code review instead.
dax@thdxr
i still don't get why we need to push code up to get an LLM review via awkward github ui hacks opencode has /review which can also do things like run your code to check things but a full time team focused on this would do it better, i just don't like the workflow they offer
English
Holy shit! We’re getting a vertical taskbar in Windows 11! And they’ve promised the impossible: fixing Explorer. Who knew it would only take five years of begging?
Pavan Davuluri@pavandavuluri
The team and I have spent the past several months analyzing feedback from the community. What came through was the voice of people who care deeply about Windows and want it to be better. Read this blog post to learn more about what we're doing in response as we look to raise the bar on Windows 11 quality. Please keep the feedback coming, to help us shape the future of Windows together. blogs.windows.com/windows-inside…
English
I think this is true.
Whether 5, 10, or 20 years, not sure, but I definitely worry very little about putting aside for pension. By that time it will mean nothing.
Rand@rand_longevity
money will not exist in 10 years, stop worrying about it
English
I use Mac and Windows interchangeably every day. When I paste something into Claude’s TUI, I’m often convinced I’m pasting the right thing. But in reality, I frequently mix up Mac and Windows shortcuts (Cmd+C/Cmd+V and Ctrl+C/Ctrl+V). When you add prompt templates to the mix, it becomes difficult to spot the error.
English
I actually like this.
I know exactly what I’m pasting so I don’t want to litter the terminal with all of that text.
However, I’d be nice to have a way to expand the text if you want to see it.
Paul Razvan Berg@PaulRBerg
This is the most annoying thing in Claude Code. Hiding raw text when you paste more than 4 lines. Terrible UX decision.
English
@trq212 I use Claude for hours every single day, and I’m still probably only using 5% of what this article recommends. The paradox is that I still think I’m at the top of my Claude game.
English
I think @EnoReyes had a point when he spoke about the importance of automated validation. We will all be spending a lot of time this year thinking about it and working on every aspect of it. And if any technology we currently use does not easily allow for automated validation, it will soon fade away. youtube.com/watch?v=ShuJ_C…
YouTube
English
@curiouswavefn I’ve seen it only with opus, and in my case, it often happens right after I stop it mid-implementation and start yelling, “STOP!!! Why would you even do X? I’ve never asked for X!”. I take it as a sign that I failed to explain what I wanted clearly enough in the first place.
English
Perhaps the most fascinating part of my LLM experience is when the model says "I am overthinking this". I then see it going back, rethinking and settling on the right strategy. Doesn't happen too often, but when it does it feels like a door that I thought was locked just opened.
English
@DmytroKrasun @yongfook But you are active in this app. How do you stop yourself from scrolling "for just a few minutes"?
English
@yongfook To stop doomscrolling at all, try to reduce the refresh rate to 0Hz.
English
Turned off 120hz back to “normal” 60hz.
Almost immediate reduction in doomscrolling.
Jon Yongfook@yongfook
Since upgrading my phone I notice myself using it more (“doomscrolling”) and I think it’s down to the 120hz screen. Everything is just faster and more fluid which activates my lizard brain and makes me go mmm just one more scroll. Concerning.
English
One key thing: tell the ai to do 2 rounds of code review on whatever it just coded, it will find most of the bugs right away. Also use coderabbit tool. I made a custom script to do it locally and track what has been scanned using git hashes of the files, run by AI during the workflow.
English
"Take screenshots and videos of everything you've built, then review it."
The amount of time this saves is wild, add onto it automated QA and you have yourself a semi-autonomous system.
I was given this advice by Leo from Factory on my podcast. youtu.be/hP5kBYbM9hU?si…
YouTube
English
I hit the limits much faster when I have a task that Claude can naturally scale across multiple subagents (10 or more). For example, localizing a large, old codebase or manipulating the contents of a large database. I almost never hit the limits when I work on the new code (in my case it is usually C# and Typescript)
English
I have long been wondering why I almost never hit Claude Code limits,
Am I writing better prompts? Using fewer agents?
Seems like the answer is that Ruby On Rails is the most efficient language for AI agents.
Irina Nazarova@inazarova
Which language is the fastest, the cheapest and which one minimizes LOC, when building with Claude Code? The answer is Ruby, and it’s winning here in every test and category.
English
@krzyzanowskim I guess I now feel a little less guilty about swearing at AI. I’m not alone.
English
tension intensifies as I keep prompting something I would fix by hand in 10 minutes.
English
if you ask me, this thing is AGI. God give me strength, because I have no more patience
English
@GregorySchier Using Yaak after Postman is like a breath of fresh air. It would be incredibly sad if this product disappeared.
English
Agentic coding will kill Yaak eventually. I saw the first proof of this today.
I've been spinning on this thought for a while. I love the project but it's extremely demotivating to work on something you know is in a shrinking market.
English
@aekae8888 @GergelyOrosz Solved with the model or the harness?
English
@fav83com @GergelyOrosz Might be more of a phase than an era. Hard to imagine the problem not being solved by 2027
English
On one end, the Anthropic team is a massive user of AI to write code (80%+ of all code deployed is written by Claude Code). They ship amazingly fast.
On the other hand, seeing these beyond terrible reliability numbers suggests there might be a downside to all this speed:
English
@GigaHiveDigital @GergelyOrosz Unless this is a medical software...
English
@fav83com @GergelyOrosz This is probably the next shift. We spent years ensuring PROD was “never broken”. Now it seems PROD issues are normailzed because Business and Product are fine with reliability issues as long as Eng can react with patches. Who knows where this goes….
English
@BenSammons @GergelyOrosz Agreed. I think we just have to get used to this. I just hope that this approach won’t be used in medical software.
English
@fav83com @GergelyOrosz I think the 9’s are providing diminishing returns in comparison to shipping. Hate it but I think the math works out. Only core providers with tons of quality guardrails will keep any semblance of reliability.
English
