Boyan Dimitrov

A decade after the West gave up on building the super app, it’s coming back- and this time it might actually work.

@lydiahallie Making burn always visible in Claude Code itself would help a lot. I built cc-budget as a proof of concept for per-prompt burn, pacing and peak/off-peak to avoid surprises: github.com/boyand/cc-budg…

Digging into reports, most of the fastest burn came down to a few token-heavy patterns. Some tips: • Sonnet 4.6 is the better default on Pro. Opus burns roughly twice as fast. Switch at session start. • Lower the effort level or turn off extended thinking when you don't need deep reasoning. Switch at session start. • Start fresh instead of resuming large sessions that have been idle ~1h • Cap your context window, long sessions cost more CLAUDE_CODE_AUTO_COMPACT_WINDOW=200000 We're rolling out more efficiency improvements, make sure you're on the latest version. If a small session is still eating a huge chunk of your limit in a way that seems unreasonable, run /feedback and we'll investigate

Thank you to everyone who spent time sending us feedback and reports. We've investigated and we're sorry this has been a bad experience. Here's what we found:

@svpino Has been the case since end of last week. Built even a quick plugin to measure the burn across sessions. Without doing anything crazy fills up super quick. 5h session limits on the Pro are hit in minutes.

I'm on a MAX plan, and now I'm hitting my limit every couple of hours. What's going on? Honestly, this makes Claude unusable. I've had to sign up for ChatGPT again just because of this.

@hoblon Nice one! :)

@nathariel Isn't it beautiful? ^_^

With the latest Claude Code session limit changes I started hitting the wall way more often - especially during peak hours I didn't even know existed. So I built a small status line tool that keeps your budget visible at all times. Pacing, per-prompt cost, peak detection, and a warning before you burn through it. github.com/boyand/cc-budg…

@afnfyz @lydiahallie Added quick support for vscode + optional notifications for you :)

@nathariel @lydiahallie Very cool, come across anything that would bring this into the vs code extension?

We're aware people are hitting usage limits in Claude Code way faster than expected. Actively investigating, will share more when we have an update!

@afnfyz @lydiahallie No, into Claude code itself: github.com/boyand/cc-budg…

@nathariel @lydiahallie Is this setup in the vscode ide?

cc-budget update — enterprise support session cost, per-prompt delta, daily + monthly spend and a discount % config if you have a negotiated rate animated demo now shows both modes 👇

@_chenglou Rarely get excited but this is some sick work, @cheng! Built a typographic Tetris demo with pretext and it is crazy how text reflows around falling blocks at 60fps using layoutNextLine.

My dear front-end developers (and anyone who’s interested in the future of interfaces): I have crawled through depths of hell to bring you, for the foreseeable years, one of the more important foundational pieces of UI engineering (if not in implementation then certainly at least in concept): Fast, accurate and comprehensive userland text measurement algorithm in pure TypeScript, usable for laying out entire web pages without CSS, bypassing DOM measurements and reflow

Install: claude plugin add github:boyand/codex-review Requires Go 1.22+ for the engine.

How it works /codex-review:plan — Codex reviews your Claude plan /codex-review:impl — Codex reviews implementation against the approved plan /codex-review:summary — severity counts + per-finding decisions Each Claude session gets its own workflow. Prior rounds are preserved.

Lately I get the best results by chaining Claude and Codex together. Claude plans. Codex reviews. Claude implements. Codex reviews again. Tried different solutions but never got to a reliable unattended mode so ended up building a plugin that makes the loop repeatable. github.com/boyand/codex-r…

@tonysheng You have my vote for feature of the quarter

I built a Claude Code notification system that uses Warcraft III Peon voice lines. It's probably the stupidest thing I've ever shipped. And according to everybody that has used it, it's also incredibly useful. (sound on)

Funny first days of 2026 with Claude Code. At first, one started maxing out the MAX 20x within 1-2 days for what took a whole week before. Now the Usage tracking is completely gone. Vibe++

More on our AI Design to Code pipeline and our platform engineering approach: slideshare.net/slideshow/how-… youtube.com/watch?v=OtxWxk…

For the first time publicly, we unveiled the core foundations that enable us at SIXT to generate ready-to-ship, full-fidelity, enterprise-quality screens directly from Figma designs. The magic happens when years of platform investments compound: Our AI-powered pipeline now transforms designs from intent into production-ready code, supporting 800+ engineers, product owners, designers and data scientists across SIXT's ecosystem. We are just getting started. If you want to join a wild journey, DM me.

Albeit being bullish on AI, I can’t stop thinking about the mountains of e-garbage being produced. All those “AI companions” creating your motivation letter and applying in your name to hundreds of jobs. And then we have “AI screening tooling” and “AI recruiters” filtering out the thrash. So much capital wasted to get back to square 1.

A fantastic retrospective from #AWS Sr. Principal Engineer @molson, who's been at the heart of scaling #EBS for over a decade. Marc shares hard-won lessons on driving continuous performance improvements - a must-read for cloud infrastructure nerds. allthingsdistributed.com/2024/08/contin…