On Dungeon Crawler Carl #5 now after starting the series in December. I now want to make my LLM speak in the style of the AI in DCC and slowly go primal, would be a blast

2000 likes and I'll build warcraftcn/ui

Reading AI PRs and AI reviews feels icky. Like no one would actually write and format stuff like this, I think the problem is it doesn’t feel like two humans chatting but chatting with an AI, but through a human where I now have 0 trust you checked the AIs work.

I found this out, 3 years ago I got limited so fast on every book for being a sharp, especially arbing stands out for its weird bets. Plus EV betting worked for a while but they limited there too. Honestly probably a good thing, got out while the gettin was good

This is meant to be cool but isn’t it also an indictment on how many resources we use to service trivial information? Not calling out v0 here, any AI is like this. I can’t help but wonder if all of this has made humanity better.

Ralph Wiggum is the craziest thing to happen in the coding agents space in 2026 so far not the technique itself. just the fact that someone was like "here's a technique it's called the Ralph Wiggum technique because of the Simpsons guy" and everyone was like "OK sounds good"

Before AI: programmers spent Christmas break rewriting their blog in new tech stack, first post is about how you rewrote it 2025 AI: blog rewrite took an hour, first post is to write about how you vibe coded the new blog, and now there are many more days to actually write blogs

Pluribus was awesome 👏

This is the AlphaGo moment for software agents. Every AI coding agent today, from Cursor to Devin to Claude Code, learns from human traces. GitHub issues. Pull requests. Test suites written by developers. Someone had to write all of it first. This creates a hard ceiling.

A few things I've noticed as all devs write code with AI. When you write foundational / architectural code of a new project by hand, you "feel" the code pushing back if your abstraction isn't right. You feel when something is harder than it should be. The code is telling you

📣 TanStack AI Alpha is here! ✨ Framework agnostic 🤖 Provider agnostic 🧠 Type safe 🔧 Isomorphic tools 🛠 Devtools 🌐 Open protocol 📦 JS, Python, PHP ⚛️ React, Solid, Vanilla 🌀 OpenAI, Anthropic, Gemini, Ollama, ++ Official blog post: 🔗⬇︎🧵

GitHubs React frontend rewrite has produced a worse frontend than before. The old Rails Hotwire approach was so snappy. I also notice so many more UI bugs and failures to be mobile responsive. I wonder if someday I will find myself on Codeberg or SourceHut

I spent the last 60 days working at Cursor. It's been one of the most thrilling phases of my professional life. There's a lot of mystique around the company. Over the last two months, some things matched my expectations; many did not. I wrote an essay for @joincolossus

In before No AI Fridays is a company perk where devs are allowed to hand-roll code once a week.

This is @RepoPrompt 1.5 The new Context Builder connects to your agent of choice, using your existing subscriptions, to fully automate the process of building the perfect context for a given token budget. All of Repo Prompt's power, fully automated, with 1 click

Liking claude sonnet 4.5, mainly if a model is steerable, correct a decent amount of time, and is fast enough, it makes up for not being the most intelligent.

In our investigation, we uncovered three separate bugs. They were partly overlapping, making diagnosis even trickier. We've now resolved all three bugs and written a technical report on what happened, which you can find here:

👀 If you thought Opus was a bit mid lately Anthropic tried a new inference stack and that caused degraded performance https://t.co/kOSfRtb5lA

neat!

It would be amazing if we could configure MCP server tools to only be injected into Claude Code subagents, that way you could isolate context for heavy mcp servers (10k+ tokens of context) like Playwright or Notion into a subagent and not pollute the main context window.