M. Adel Alhashemi

@robinebers @b1rkh0ff It really will not, you clearly haven't used a proper CLI orchestration layer.

@b1rkh0ff this reply will age like good milk

hot take: Cursor's real endgame isn't competing with Claude Code or Codex it's becoming their BOSS the moment Cursor ships proper subagent orchestration (weeks away at best) on top of the skills, it becomes the meta-layer on top of every AI coding tool seamlessly Cursor can use ALL of them picture this: Cursor becomes the orchestration layer need deep reasoning? orchestrates Codex need fast building/research? orchestrates Claude Code need specialized tasks? think: - in-built browser - debug mode - best-of-n / multi-model asks - models unsupported by claude/codex do it right inside of Cursor instead with their inference Cursor doesn't have to beat them it can COMMAND them because they can compete on a lot of things, but tokens isn't one of them that's not an IDE anymore - that's an AI workforce manager TL;DR: - use Cursor for all the goodness they offer - orchestrate insane token tasks to Claude/Codex - virtually zero downside for users, and no way to prevent it either this is the future, and I genuinely think I saw it this morning

@theo Hey how about you make t3.chat the OS of chatting apps haha. I mean why can't I use my 200 OpenAI sub that I am using with OC in a better interface like t3 chat? Would be nice of OpenAI to show their good will here as well

So I kind of want to make an operating system…

❤️

@gdb This is how you wind the dev community! Kudos

@thdxr Surely I am, a great outcome honestly.

we are working with openai to allow codex users to benefit from their subscription directly within OpenCode

@thsottiaux So just to be clear, are you guys totally ok with us using Codex sub with OpenCode without us being considered as breaking your ToS? If true I am giving you my 200 USD instead of Anthropic

With the Claude Code shutdown. I am proud that we build Codex in the open with our OSS repo and we are 100% invested in supporting a flourishing ecosystem of agentic coding tools out there. You can already build on top of github.com/openai/codex/t… directly, which includes ChatGPT login and same usage as you get in codex. Reach out if you are a builder!

@oscabriel @thdxr You lucky 🤬

@thdxr I got in

it sold out in 1 minute i literally thought someone scalped them all but it's all different people we'll be back next week

@thdxr More like in 36 seconds. How many seats were you offering?

@opencode Couldn't pay, joined within 26 seconds of you posting, does not make any sense

buy.stripe.com/cNi28k0crfI1fv… we're still activating people manually expect an email later today

another batch of opencode black is now available $200/month use any model generous limits link to buy in reply

@opencode You are late by 1 minute

tomorrow 12pm EST

@erencemay @thdxr Me as well, the most reasonable move

on the anthropic stuff (technical details in a reply) it's their business they have the right to enforce their terms however they like they're not obligated to provide completely open access to their services but it is a hint at the underlying problem - models aren't sticky so it's not good enough that you're using opus they need you and your team to adopt their full stack of tooling so that it is hard to switch to a better model thankfully not all LLM providers want to operate this way - you'll hear more about that soon :) in the meantime we already teased more OpenCode Black dropping today - so stay tuned

@thdxr "thankfully not all LLM providers want to operate this way - you'll hear more about that soon:)" I wonder who that is 🤔 @grok is it you?

@thdxr I believe more providers will go their route in the future unfortunately, I am willing to bet the day will come that some coding specific models will only be provided through their subs and not directly through the API.

@nummanali @bcherny The only problem I have with output-styles is that they don't have custom configs, the way I see it they enable use cases outside of coding and it would be way more useful if you can define what MCPs skills etc each output style has access to.

Claude Code Output Styles Highly recommended level up for all experience level Using /config set your output style to one of: - Default (focused on coding, no fluff) - Explanatory (product thinking, concepts, debugging) - Learning (guides you to write code - amazing!)

sigh cc letting skills get activated with a slash command which has massive overlap with custom slash commands this is why i'm reluctant to match every cc feature - i care a lot about doing good product work i don't want opencode to become a mess

@thdxr As a heavy skills user I see the strong value in skills, however I like what we currently have, commands are user invoked and skills are agent invoked. Tool permissions on agent level, I really think OpenCode should keep things clean this way, makes so much sense.

@thdxr Oh this is not the whole party, skill tool can now let the agent call slash commands, and you have to opt-out of this in the slash command frontmatter.

i can't even keep track of this shit what is the difference? what does activating a skill even mean, do you have to send a prompt with it? skills can be have limited tools? what? what happens if you continue the convo

@thdxr WHY THE HELLL DID I HAVE TO SLEEP 😭

sold out in an hour for people asking about rate limits - the goal here is to provide a similar offering to claude max/codex pro and we're using this initial batch to understand usage patterns then we'll make it more widely available with cheaper plans as well

@skeptrune Now show me an image why would you use OpenCode over Claude Code? 🙃

here's my explanation for why people use Claude Code instead of Cursor

@theo Man I have been looking for such a tool for so long that I almost gave up and was about to build one. Thanks for sharing!!!

Master class is getting my attention. This is cool as hell