Ma Ga Re

@Zortosdev @cursor_ai @OpenAI How did you decide which prompt we should use for both of the models?

I tested @cursor_ai Composer 2.5 vs @OpenAI GPT-5.5 Medium Fast in a head-to-head coding speed test. Same prompt. Same task. Results: Composer 2.5: 43 sec GPT-5.5: 2 min 3 sec Composer finished faster, while GPT-5.5 took longer and made more tool calls.

@paulg I always select the first option.

If a site or app forces me to answer some unnecessary question before I can continue, I always give a wrong answer. How did you hear about us? Radio. Does anyone else do that?

@OjasSharma276 Companies will shift to cursor where composer is dirt-chip. Babu companies will die.

From 1st June, people who rely on AI for everything in their companies are gonna get cooked. In our stand up, we discussed that we can’t use AI for every single task anymore. The token consumption for each model is getting way higher. Right now, Claude Opus 4.6 takes 3x tokens, but from 1st June it’s apparently going to take 27x. Your GitHub Copilot quota will probably be gone within the first 2-3 days itself.

@jh3yy Create a screen studio like Recorder Chrome extension. That is challenging.

might build a screen recorder 🥁

@nomad_suzuki Is that the chair from The Conjuring ?

3面LEDの上手な表現

@rachelnocode Does it support multiple accounts for the same provider?

are you still noticing problems with Codex limits?

@appfactory codex is Apple, Claude is linux and Gemini is Microsoft

Codex is Apple. Claude is Microsoft.

@pkuhar that's fucking awesome

How do you know your agent is still running after you close the lid? I became a bit obsessed with putting some RGB LEDs for a status indicator into the SD Card slot on my MacBook Pro. I never use that slot, so it's a perfect place.

@jturntdev My 5x plan is looking like a 5x plan. I've stopped using /goal.

Codex Usage Limits was fine for two days, now cooked. Either， /goals is completely broken and drains usage 2-3x faster. Or their caching has destabilised. Im on 20x and it feels like im on 5x

@Hiteshdotcom I am building my own which will have some cinematic features.

Market is flooded with screen recording software with auto zoom. Simple pattern, add zoom on every click. Nothing new. And sooo many of them. From open source to free, free 5$ to lifetime plans, some are available to acquire. But all feels same. And I have tried 10+ so far.

@nikunj You can just navigate the website. Download the network request as HAR and ask Claude Code to reverse-engineer it. Browser harness is too slow in this case.

Reverse engineering APIs through network requests is one of the most fun things you can do with Claude Code to automate tasks.. SO many websites are impossible to navigate "deterministically" via the DOM (or through screenshots). So, I just point Claude Code to use browser_harness by @browser_use (or vanilla playwright) and ask it to sniff network requests on the pages that I'm trying to get info on. And, I just keep clicking around on the sections of data that I want. And, then Claude Code is generally able to go through the logs to figure out what is the right structure for these APIs and what kind of auth do they need (most are cookie based). We also determine what kind of rate limits exist based on trial and error. I'm able to use that to construct jobs that allow me to get that data programmatically. There are many use cases for this besides scraping. I use this for random side projects (like the travel CLI), for monitoring websites (for intel), and for many many other use cases. Every website will soon need to be headless, and we'll need to figure out mechanisms for how we have our agents pay these websites programmatically as well. Just as we have llms.txt for data and structure, we'll soon need tools.txt for agents to determine what tools exist that can be leveraged.

@TonyTheMathLion How many of them have you solved?

Are they really 'among the hardest questions'???

@arnavmehta007 Bookmark this post. For now, I have the required compute. I'll contact you only if it is necessary.

@VilasMagare5 serious, yes, let me know

I'm willing to provide access to Thousands of $$$ in Compute - H100s, H200s, RTX 6000 Pro Blackwell GPUs if you are willing to open-source something legit (unlike Sarvam) for the country

@StefanFSchubert Make it at least 10.

That’s a terrible rule, have you met a five-year-old

@sheherenow_ I can't connect my Amazon Echo to figma, or do I? @grok Can I connect my Amazon Echo to figma?

@VilasMagare5 it was just an amazon echo lol

lol i found this demo video from 2017 - my team at airbnb made plausibly the first voice interface for a design tool~ the only thing worse than being too late is being too early!

@AniC_dev I love the introduction, by the way

@VilasMagare5 you need to first think inside then tear it appart

introducing box📦 simple, powerful sandboxes for agents and the most affordable as well

@zzzzshawn @rexan_wong I have the same app which does the same thing. Didn't release just because somebody made it open source and for free. I need to have the confidence of that guy.