kluster.ai

Introducing Repo Reviews! 🎉🎉🎉 Learn how to run system-wide codebase analysis with Repo Reviews 🧠 docs.kluster.ai/code-reviews/r…

@ai_for_success We like both ♥

What are you using more right now for coding? Claude Code or OpenAI Codex?

lol. Should we do real-time ad-reviews now?🤣

@SyahmiRafsan Hey @SyahmiRafsan, we can make sure to give real time feedback to Codex. Want to try?😉 kluster.ai 👈

My 24 hours experience with the Codex app. 5.2 Codex Medium is fast enough and smart enough for daily use. Only using Cursor to manually check code but most of the time I am only in Codex. Expect some bugs when accepting and reverting changes. But the biggest issue is that I cannot edit the sent message to the model (?) Which is a bit weird since OpenAI is the one introducing chat-like branching. Hate to say it but I might just use Zed for IDE (can it beat Cursor's tab completion?!) to pair with Codex app.

@mewtyunjay Sound like we can help with this. Or agent mode reviews on real time and gives you feedback and actions to fix. We are faster than Codex 😎

Claude Code/Codex writes most of my code these days and I use lazyvim/neovim to review the diffs. Even with plan mode, there are times I don't fully understand the changes being made, so I built a tool that lets me handle the entire post-coding process in one place. This is my first attempt at bridging the gap between learning and doing with the speed of AI.

AI makes writing code cheap. Understanding it? That's the expensive part now. New devs ship code they don't fully grasp. Learning frameworks from scratch feels wasteful when AI writes most of it anyway. So I built DiffX, a middle ground. Treat AI as your junior engineer. Your job is to review what it wrote. Diffs become the first-class citizen with gated permissions before anything gets pushed to enforce learning.

@feO2x Hey @feO2x should try us. We can also check your repo, or the code while it's being made. We'll love to hear your feedback: kluster.ai

💯 Achievement unlocked: GitHub Copilot reviewed a PR and did not generate any comments on how to improve the PR. #AI #GitHub #Copilot #Code #Review #dotnet #csharp

Announcing Deep Reviews🥳 Quick iteration or final review? We got you covered: ⚡ Instant: ~5 sec feedback loops 🧠 Deep: catches edge cases your tests missed Speed when you need it. Depth when it matters. Free tier: kluster.ai

@0xgaut Let us do the watching and you can think on cool features.

watching claude code work acting like you know what it’s doing

@0xgaut You can actually reduce your context window if you let us review the code that Claude generates. free tier: kluster.ai

"Claude usage limit reached. Your limit will reset at 7 AM"

@dillon_mulroy @dillon_mulroy 👀 youtube.com/watch?v=rpWt9s…

my take on the code review tools that are coming out: they’re solving real problems, but not the biggest problem: i need a way to review code effectively before it ever gets to github

@talkaboutdesign @corbtt @talkaboutdesign 💯 That's why we are here. No more PR bottlenecks. Free trial 😉

@corbtt the amount of code being shipped is going through the roof. there is simply no time to manually review everything. huge bottleneck.

It still pains me a bit to say it but the "humans should not be reviewing code" side is definitely going to win. Even if the models never get better than today (they will). You carefully review the design docs, and the testing plan. But not the generated code.

@TheAhmadOsman @TheAhmadOsman we also like to watch.👀 Instant code reviews for claude code. Try it out. -> kluster.ai

me watching Claude Code write the code for me

@thomaspaulmann Top reviewer reporting to service, sir.

Claude is now our top committer 🤯

@tunguz Who's gonna review all that code?

Rate my new keyboard.

@EvanWritesOnX @EvanWritesOnX, @Morgoth_888 Bottleneck? Gone. 💨 Want to try it out? -> kluster.ai 😎

The reason for this is simple. The bottleneck is not at the engineering level where code is written. Almost every engineer I know already writes code with AI well before Codex, or Claude Code came along. The bottleneck is at the code review level, and accountability. When Anthropic decides to push out a product called “Claude Security” and announces that the company will now take full guarantee, accountability for code reviews and security, that’s when the industry will change. Overnight.

@NaveenS16 Step 1: Install kluster.ai...😎 @NaveenS16 try us out, free tier

#AI can write code super fast. But unreviewed AI-generated code going to production is how you create quiet disasters. Here’s how to review AI code properly—and what really happens when you don’t. 🧵👇#VibeCoding #SoftwareDevelopment #SoftwareEngineering #DevOps

@fredyfredo123 @burkov We can review those lines for you. Try it out 😉

@burkov It's still not matured. The main issue in coding is the code review. We cannot review 10k++ code lines a day. Maintain a claude code is not really possible

Cursor and Perplexity are two products that were only relevant for a short moment in time when the LM tech hadn't matured yet and an army of enthusiasts had nowhere to go. Just like Mistral 7B Instruct was good when it was released but only until the tech industry started to seriously invest in open weight models. It's time to say goodbye to these two. These were a fun three years, but these pet projects aren't relevant anymore.

@ashleyrudland 👀 @ashleyrudland, let us help with reviewing the code of all your agents. kluster.ai

Yeah I think the limiting factor is the code review, you can ship slop so you are forced to quadruple check every line of code, test a LOT, which makes the cognitive load very high right now because the volume is so high. Also exciting AF 😈

@rahulj51 Hey @rahulj51 We work very hard to reduce the bugs before the PR stage. 💪 (free trials for enterprise😉) -> kluster.ai

PR reviews are the biggest bottlenecks right now. Github should allow tweaking approval-required rules to be further fine-tuned based on LOCs or specific paths etc.