🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

The github pull request extension is an absolute life saver when trying to give feedback to lots of async coding agents at the same time You can easily check out the code, run it, leave comments inline, make small fixes as you go along, then submit a review and say "@claude fix

Unfiltered first impressions of cursor background agent - At first nothing at all loaded in the UI for ages - At first glance, it seems to run on some ubuntu VM and then remotely connects the local developer UI to the remote dev server (vscode has an SSH remote mode like this

X is not down 🧐

This is the way

Okay imagine this: claude code in github CI, except when it boots up it starts a server and posts a link to it (using something like ngrok) in the issue or PR comment Click on the link to get access to claude running in an interactive terminal in the browser where I can

... one ... more ... prompt Every evening when my wife and kids go to bed I start coding thinking "only half an hour", but then it's three hours later and I've got three coding agents on the go and several in-depth deep research conversations with o3-pro scouting the problems of

After ~6 years of building these types of architectures (starting with BERT, eg see Baleen), I think calling these multi-agent systems is a distraction. This is just software. Happens to be AI software. It doesn’t seem so complicated once you internalize it’s just a program.

What if Cloudflare bought bun and merged it with workerd? I realise this makes little technical sense, but as user I would love bun’s DX on cloudflare’s platform and I’d love to be able to use the same runtime my deployed code uses when writing scripts or in my test runner etc

Nbd, just full-stack apps in an npm package Queues, storage, agents, workflows, frontend, backend, assets, the whole lot Apps can depend on other apps, send requests via RPC and publish/subscribe events via shared event bus Blank is cooking

Actually, have just gone all the way and deleted our scripts in favour of a markdown file (with some bash commands inside) Now any time I want to change engineering onboarding, I can just throw some prose into that file (e.g. "If something weird happens with node, double check

Who’s making this but across all agents I’d love to gamify this in the team 1) who uses async AI agents most effectively 2) A month later, how many of the LOC are still there Could also have breakdown to show how much is test vs business logic vs documentation, etc I guess

Equally apprehensive and excited about the naked gun remake The original is one of the 5 best movies ever made

My hot take on coworkers: Great for monitoring PRs and GitHub issues but I don’t want them to build features I kinda enjoy pairing up with coworkers and building something cool What am I gonna do if coworkers code everything??

if you get good at systems design and then use cursor to build you can create incredibly robust products. this is the opposite of vibe coding and still leverages AI at full power.

Some kind of “chorus for async coding agents” would maybe be cool My preference would be for it to be hooked up to linear. I assign the coding meta agent bot to a task, then it spins up a separate thread (right there on the linear ticket!) for each coding agent and tells them

Claude opus gladly copy a 500 LOC file token for token rather than invoke mv