@gregisenberg I want my remote session with Codex or Claude to have a live voice conversation with me about what's going on. Video is a bonus I can wait for, even.
Limiting this to generic chat instead of work going on in my actual local projects is the real gap.
Really impressed with OpenAI's Codex for inter-chat communication automation.
Saves me from copying a prompt from one thread and pasting it into another, so two parallel threads can talk to one another without starting as part of a planned orchestration or coordination.
@theo Does that mean Claude slash commands and Claude Code hooks are respected (i.e. leveraged by the GPT-5.x model), or is it just the raw prompt/response UI from the OpenAI model that the is proxying through?
If you have YouTube Premium, you may find as I have, that the new 'Claude FM' video stream serves as a decent, commercial free alternative to a https://t.co/PX011ymkm7 subscription.
Take a listen to Claude FM here: https://t.co/Y7hNHpDTXX
@thinkwithmark For now, I will use a pen to underline and then go over it a second time with Mark™. Maybe there's a way to integrate an optional physical highlight at the same time as the digital capture? /2
@thinkwithmark Looks cool. I would love if this integrates with Readwise, too. I still like using a colored highlighter or physically underlining my book so I can see the highlight in context. /1
AI is raising the bar for technology professionals—but not only on hard skills. Brian Elliott and Dave Blundin make the point well:
As AI helps more people do technical work faster, soft skills become the edge. Right now, they’re a real differentiator, especially in AI.
@pxue Excellent take. I agree with you, Paul (and we're fellow alums, it turns out). I too, want to see us build Canada strong and for our incredible talent to build a thriving technology sector here.
@ryancarson I earmarked today for exactly this; then read your tweet! In my case, I've wanted to clean up and systematize conventions, naming, and docs for agents.
Will require severing the link to upstream for useful skills I've gathered from the web so I can standardize how they work.
We're all connected and there doesn't seem to be enough capacity or resilience to redistribute load/services seamlessly when a datacenter or two is impacted.
Claude Code status line setup that includes billing mode so you see '✅ Max' if your subscription is being used, and '⚠️ API' if it is using token based billing.
https://t.co/dqBhXgvuXQ
Going to watch this soon. Hopefully, you discussed a way to put this Mac app in YOLO mode. The permissions pestering has been so annoying. I'm normally a Claude Code user; so I must be missing something.
YOLO mode for Codex CLI is easy/clear; but it's a mystery to me thus far with this Codex Mac app.
@OpenAI I'm pestered with approvals. How do we configure this to run 'YOLO'?
I can run the Codex CLI in yolo mode, but the Mac app, while inviting and promising, is just unusable with all the permission pestering.
These Anthropic ads dissing advertising inside AI products are just killer. Kudos to the marketing team at Anthropic who put these together!
https://t.co/9lMyXJwc6t
I'm enjoying building with Rails and Claude Code for these very reasons. And also because human review of PRs, when I want to to review the code, is less cognitively burdensome, given the conventions and patterns set out by the Ruby on Rails framework.
I think people are sleeping a bit on how much Ruby on Rails + Claude Code is a *crazy unlock* - I mean Rails was designed for people who love syntactic sugar, and LLMs are sugar fiends.
@dwlz I'm getting the same, even after a fresh reboot and a fresh /login to my Max account:
API Error: 500 {"type":"error","error":{"type":"api_error","message":"Internal server error"},"request_id":"req_011CXmQ2rQ5mPrjyU4sHENMN"}
Good principles. I've had Claude Code create me a slash command I can invoke that refactors https://t.co/B2CylFQlx6 contents into topical documents, pointing to each one, just as the article suggests structuring it. Helpful to re-run whenever new stuff gets appended to https://t.co/B2CylFQlx6.