@kr0der Anthony
Anthony posts on X about ai, vercel, super, mode the most. They currently have [-----] followers and [---] posts still getting attention that total [-----] engagements in the last [--] hours.
Engagements: [-----] #
- [--] Week [-------] +129%
- [--] Month [-------] +9%
Mentions: [--] #
Followers: [-----] #
- [--] Week [-----] +14%
- [--] Month [-----] +34%
CreatorRank: [-------] #
Social Influence
Social category influence technology brands 7.74% social networks 3.87% finance 2.58% stocks 0.65%
Social topic influence ai 5.16%, vercel #171, super 3.23%, mode 3.23%, code 2.58%, slack #218, claude code 2.58%, cursor 2.58%, open ai 1.94%, anthropic #1491
Top accounts mentioned or mentioned by @llmjunky @tejashaveridev @cristianxiv @solaawodiya @ernielm @rudrank @robinebers @steipete @danielwhit21874 @embirico @anasibnanwar @jarrodwatts @bil0090 @jediahkatz @realhasanshoaib @udaysy @thsottiaux @rhyssullivan @fredrikaurdal @rynhay
Top assets mentioned Cloudflare, Inc. (NET)
Top Social Posts
Top posts by engagements in the last [--] hours
"@thsottiaux used this to verify everything is all good thanks to alexander 🫡 https://x.com/embirico/status/2021643604650078482s=20 https://x.com/embirico/status/2021643604650078482s=20"
X Link 2026-02-11T17:58Z [----] followers, 12.1K engagements
"great point by Simon - i noticed this with my own side projects sometimes where i have no idea whats in the repo anymore so its hard to make changes easily Short musings on "cognitive debt" - I'm seeing this in my own work where excessive unreviewed AI-generated code leads me to lose a firm mental model of what I've built which then makes it harder to confidently make future decisions https://t.co/KUqQXDVNiS Short musings on "cognitive debt" - I'm seeing this in my own work where excessive unreviewed AI-generated code leads me to lose a firm mental model of what I've built which then makes it"
X Link 2026-02-15T19:05Z [----] followers, [---] engagements
"@LLMJunky feels wrong to have to submit your ID due to a bug that's been going for days already"
X Link 2026-02-16T05:09Z [----] followers, [---] engagements
"Codex app has a bug - i have "never" for approval and "full access" sandbox mode but it asked for approval to run "npm i""
X Link 2026-02-03T05:16Z [----] followers, [----] engagements
"using the Codex model in the Codex CLI in the terminal of the Codex app which is a UI for managing multiple Codex chats running Codex models that code and create PRs via the Codex app which get reviewed by the Codex PR reviewer where you tag Codex cloud to fix it which uses the Codex model to fix it and you tag Codex to review again then you use the Codex app to merge the PR and you ask Codex to update the changelog https://twitter.com/i/web/status/2018639035577823447 https://twitter.com/i/web/status/2018639035577823447"
X Link 2026-02-03T10:53Z [----] followers, [---] engagements
"alright it doesn't exist yet i also tried Composer [--] 🤣 that's it i'm getting access to Sonnet [--] early via Cursor https://t.co/cqMPqTsLhV that's it i'm getting access to Sonnet [--] early via Cursor https://t.co/cqMPqTsLhV"
X Link 2026-02-04T02:53Z [----] followers, [----] engagements
"i noticed Codex (the model) is so good at following instructions that it'll just clean up random code as it goes along just because it adheres to the instructions or your AGENTS md. Opus just completes its task and calls it a day as if it's watching the clock and ready to leave work at [--] pm. both are great models and this isn't a major upside/downside just an interesting observation https://twitter.com/i/web/status/2018981490760319070 https://twitter.com/i/web/status/2018981490760319070"
X Link 2026-02-04T09:34Z [----] followers, 21.3K engagements
"massive improvement to token efficiency bro what codex [---] was already insanely token efficient that 278k tokens felt like 1m already. also codex [---] is 25% faster gg lets go Straight from the oven from the OpenAI naming team GPT-5.3-Codex is here. Update to latest version of Codex App or CLI to enjoy it. It's a massive improvement on token-efficiency and on top of this we are running on an improved infrastructure and inference path that makes it Straight from the oven from the OpenAI naming team GPT-5.3-Codex is here. Update to latest version of Codex App or CLI to enjoy it. It's a massive"
X Link 2026-02-05T18:18Z [----] followers, [---] engagements
"despite Anthropic getting hate for gatekeeping CC subs recently under safety OpenAI is actually the one doing this as new Codex models are gatekept to the Codex CLI. at least you can use the new Opus models via API whereas Codex [---] you HAVE to buy a Codex sub. i love Codex as its the best coding model currently but this really sucks that youre forced to get a sub to use it. Why did they "release" codex [---] yesterday but its not in cursor today while claude opus [---] is 🤔🤔 Why did they "release" codex [---] yesterday but its not in cursor today while claude opus [---] is 🤔🤔"
X Link 2026-02-07T06:33Z [----] followers, [----] engagements
"currently you need [--] subscriptions: [--]. one that allows you to use Opus for frontend (Cursor CC) [--]. Codex this will continue indefinitely until either Codex beats Opus at frontend or new Codex models arent delayed in the API by over a month"
X Link 2026-02-07T20:19Z [----] followers, [----] engagements
"@PawelJLisowski in large codebases or complex tasks youll wanna go with codex from my experience working with a 1m LOC codebase. opus rushes too much which isnt as ideal for large codebases. as a test try run codexs /review command after opus does a large task"
X Link 2026-02-08T05:29Z [----] followers, [--] engagements
"the main downside of codex 😂 it doesnt know it can do things or run tools half the time my problem with codex lol https://t.co/aceIiAWj4v my problem with codex lol https://t.co/aceIiAWj4v"
X Link 2026-02-08T10:07Z [----] followers, [---] engagements
"Anthropic just dropped their most expensive model yet - Opus [---] Fast. 6x more expensive than Opus [---] at $30/m input and $150/m output. Not sure what the use case for this is. You have used your $200 of Anthropic API credits for this message. You can enable auto-reload in the Anthropic Console. https://twitter.com/i/web/status/2020694664219865302 https://twitter.com/i/web/status/2020694664219865302"
X Link 2026-02-09T03:02Z [----] followers, [---] engagements
"Codex CLI vs Codex App: small screen (e.g. laptop) = use Codex App since it's nearly impossible to have many CLI chats open at once on laptop since the Codex chat doesn't resize when you resize your terminal big screen (e.g. monitor) = Codex CLI since you can open and see many chats at once (e.g. @steipete's workflow) https://twitter.com/i/web/status/2020741136956973367 https://twitter.com/i/web/status/2020741136956973367"
X Link 2026-02-09T06:06Z [----] followers, [---] engagements
"@rudrank @cursor_ai love their background agents for this + slack"
X Link 2026-02-09T10:32Z [----] followers, [--] engagements
"@gauravnvarma @RhysSullivan vercel domains"
X Link 2026-02-09T10:37Z [----] followers, [--] engagements
"@FredrikAurdal @steipete i didn't find claude that good at invoking skills compared to codex what's your experience"
X Link 2026-02-10T01:20Z [----] followers, [----] engagements
"goodbye Opus hello Composer 1.5"
X Link 2026-02-10T01:34Z [----] followers, 22.9K engagements
"@_bgiori @steipete codex is really good at these weird UI bugs for some reason 💀"
X Link 2026-02-10T02:38Z [----] followers, [----] engagements
"@LLMJunky @steipete bro's cleaning up after opus"
X Link 2026-02-10T04:50Z [----] followers, [----] engagements
"@jarrodwatts @cifilter [--] context window feels like it lasts ages"
X Link 2026-02-10T04:53Z [----] followers, [--] engagements
"Composer [---] shouldn't be compared to Codex [---] pricing it should be compared to Opus [---]. there are [--] categories of models one for super fast sync work and one for long running tasks. Codex [---] is clearly the best at long running tasks while Composer [---] is more in the category of super fast sync work"
X Link 2026-02-10T06:06Z [----] followers, [----] engagements
"@grimcodes i like the UI makes installing plugins way easier"
X Link 2026-02-10T06:07Z [----] followers, [---] engagements
"@steipete a further breakdown about why i swapped from opus to codex: https://kr0der.com/newsletter/why-i-swapped-from-opus-to-codex https://kr0der.com/newsletter/why-i-swapped-from-opus-to-codex"
X Link 2026-02-10T07:26Z [----] followers, [----] engagements
"@robinebers i love vercel [--] clicks to deploy just saves insane amount of time"
X Link 2026-02-10T08:25Z [----] followers, [---] engagements
"add this to your AGENTS md so that Codex can read from your Cursor rules: "Read .cursor/rules at the start of work and follow applicable rules." i do this because i use both Cursor and Codex"
X Link 2026-02-10T08:56Z [----] followers, [---] engagements
"@tarabya_me haha true im mostly doing it for fun since im expecting cursors UI to have a big update soon"
X Link 2026-02-10T10:36Z [----] followers, [---] engagements
"@DanielWhit21874 yea composer's great for quick changes i still need to test composer [---] more"
X Link 2026-02-10T11:59Z [----] followers, [--] engagements
"to be honest i didn't use it a lot today because i hit my plan limit 💀 i just bought a 2nd sub so i can test it way more in the next few days it's great for the usual use case of asking codebase questions + quick edits + subagents though i tested it against gpt-5.2-high for the same plan and the plan wasn't bad actually but a little worse than gpt's plan https://twitter.com/i/web/status/2021193306907214087 https://twitter.com/i/web/status/2021193306907214087"
X Link 2026-02-10T12:03Z [----] followers, [---] engagements
"a big cursor update coming before the end of feb maybe an agent orchestrator UI @nikku696969 Big one before the end of the month :) @nikku696969 Big one before the end of the month :)"
X Link 2026-02-10T20:59Z [----] followers, [----] engagements
"Cursor Agent Review suggestion: new option to "run before committing and cancel the commit if issues are found" sort of like a pre-commit hook. you can think of it like a "looks good to me but double check before committing""
X Link 2026-02-10T21:05Z [----] followers, [----] engagements
"@chris_karani if it becomes somewhat useful to me then i will"
X Link 2026-02-10T21:17Z [----] followers, [--] engagements
"@CristianXIV oh true good point"
X Link 2026-02-10T21:32Z [----] followers, [--] engagements
"@SIGKITTEN "hmm i see some mentions of the word clanker in your codex chat logs come here""
X Link 2026-02-10T21:35Z [----] followers, [--] engagements
"@CristianXIV that'd be good opus already uses $5 per task in [--] minutes anyway 💀"
X Link 2026-02-10T21:36Z [----] followers, [--] engagements
"@CristianXIV you can buy a second sub if needed 🫡 my free usage hasn't run out yet for some reason"
X Link 2026-02-10T21:48Z [----] followers, [--] engagements
"how high is this free usage gonna go 💀"
X Link 2026-02-10T21:48Z [----] followers, 13.5K engagements
"@velocitycake bro's using up the whole codex plan in one go"
X Link 2026-02-11T05:30Z [----] followers, [---] engagements
"@xuguangqi_ming let me know how it goes"
X Link 2026-02-11T05:30Z [----] followers, [---] engagements
"@MoreCoinCrypto oh interesting when i get codex to review its own work it often finds bugs in what it did (similar to a human making mistakes and finding bugs)"
X Link 2026-02-11T06:45Z [----] followers, [---] engagements
"alright this was the limit 🤣 time to go second account"
X Link 2026-02-11T06:46Z [----] followers, [----] engagements
"@RhysSullivan codex [---] has a plan mode in the codex CLI idk if it's readily available yet but i have this in my /.codex/config.toml under features collaboration_modes = true"
X Link 2026-02-11T06:50Z [----] followers, [---] engagements
"hypothetically if you're on a $20 plan and get $100 free usage (no idea if this is the case) and you hypothetically buy [--] cursor accounts don't you hypothetically pay $200 for $1200 of usage alright this was the limit 🤣 time to go second account https://t.co/J0ajUAdQZ4 alright this was the limit 🤣 time to go second account https://t.co/J0ajUAdQZ4"
X Link 2026-02-11T07:00Z [----] followers, [----] engagements
"@robinebers true that. at the same time you can also see which have adapted AI TOO much e.g. Slack activity tab has been broken for months but theyre still shipping new AI features"
X Link 2026-02-11T08:26Z [----] followers, [--] engagements
"@unnamed1tw @DanielWhit21874 i actually used up the pro account with [---] codex because i was doing something hard. ill probably upgrade to the next plan tier to test out composer more"
X Link 2026-02-11T08:27Z [----] followers, [--] engagements
"@kennedybaird_ its better and cheaper than opus [---] but sucks at frontend so keep using both imo"
X Link 2026-02-11T08:29Z [----] followers, [---] engagements
"@ephraimduncan i don't wanna use claude code i already pair cursor with a codex sub"
X Link 2026-02-11T09:42Z [----] followers, [---] engagements
"@rynhay oh yeah i have one alongside a cursor sub i use both"
X Link 2026-02-11T09:43Z [----] followers, [---] engagements
"@MomentodeFala codex spawns its own"
X Link 2026-02-11T09:47Z [----] followers, [--] engagements
"@LLMJunky huge also looks really easy to use. i remember trying /statusline in claude code and it just permanently hallucinated about what it could add to the status line and kept calculating context incorrectly 💀"
X Link 2026-02-11T10:13Z [----] followers, [---] engagements
"@rudrank @MiniMax_AI sheesh minimax 2.5"
X Link 2026-02-11T11:44Z [----] followers, [---] engagements
"@dom_scholz @aidenybai thats genius"
X Link 2026-02-11T11:46Z [----] followers, [--] engagements
"@BNBGer not the subs but you can use those models within cursor. i have both a cursor and codex sub"
X Link 2026-02-11T11:47Z [----] followers, [--] engagements
"@Daeshawn probably i dont use the codex app"
X Link 2026-02-11T12:38Z [----] followers, [--] engagements
"@rudrank @MiniMax_AI just saw this let's gooo minimax [---] opus 👀"
X Link 2026-02-11T12:57Z [----] followers, [--] engagements
"a completely normal prompt just got blocked in Codex. first time ever seeing this and i was trying to do real work on a codebase. i hope the false flags don't start appearing that often cc @thsottiaux"
X Link 2026-02-11T15:18Z [----] followers, [----] engagements
"@initjean on which plan 💀"
X Link 2026-02-11T15:19Z [----] followers, [--] engagements
"@iamsahaj_xyz @reach_vb agreed you can already rewind in the codex CLI by double tapping escape and it's 100x"
X Link 2026-02-11T15:36Z [----] followers, [--] engagements
"recommended watch from @theo - it covers a huge problem with agentic coding currently especially with working on multiple projects at once. needing multiple IDEs/terminals/dev servers/browsers open and not easily knowing what project you're tabbing into and it also being messy moving between projects. i'm experiencing this issue too so whichever tool solves this i'm gonna use it instantly if it feels good to use https://www.youtube.com/watchv=YVq28OTPCKw https://www.youtube.com/watchv=YVq28OTPCKw"
X Link 2026-02-11T17:10Z [----] followers, [----] engagements
"does this mean i'm being routed to GPT [---] currently since i got flagged by mistake i get that they're fixing the transparency here but this is one of the worst user experiences i've ever had with an AI product. a completely normal prompt just got blocked in Codex. first time ever seeing this and i was trying to do real work on a codebase. i hope the false flags don't start appearing that often cc @thsottiaux https://t.co/39XyFMOihA a completely normal prompt just got blocked in Codex. first time ever seeing this and i was trying to do real work on a codebase. i hope the false flags don't"
X Link 2026-02-11T17:21Z [----] followers, [--] engagements
"@embirico @murchiston @Angaisb_ it's still overflagging i just encountered it on a completely normal prompt. am i being rerouted now there's [--] transparency in the UI"
X Link 2026-02-11T17:49Z [----] followers, [---] engagements
"@embirico @murchiston @Angaisb_ nice thank you it says [---] codex still 🫡 thanks for the quick response"
X Link 2026-02-11T17:58Z [----] followers, [---] engagements
"@embirico good point always forget codex is open source 💀"
X Link 2026-02-11T18:13Z [----] followers, [---] engagements
"i unironically typed in a prompt in plan mode in Cursor with Composer [---] alt tabbed for [--] seconds and came back and the plan was fully written thought i was tripping 💀"
X Link 2026-02-11T19:06Z [----] followers, [----] engagements
"great article on how to use Codex effectively this one's a must-read. it goes into how they use Codex to create software with [--] lines of human written code. 📣 Shipping software with Codex without touching code. Heres how a small team steering Codex opened and merged [----] pull requests to deliver a product used by hundreds of internal users with zero manual coding. https://t.co/2GaeX7We2n 📣 Shipping software with Codex without touching code. Heres how a small team steering Codex opened and merged [----] pull requests to deliver a product used by hundreds of internal users with zero manual"
X Link 2026-02-11T20:05Z [----] followers, 72.6K engagements
"does anyone else like that Codex [---] is slow it feels relaxing running multiple chats/projects in parallel since you know you can go do something else while waiting and you know you're gonna come back to a [--] shotted task (assuming your prompt was good)"
X Link 2026-02-11T21:05Z [----] followers, [----] engagements
"@rudrank pro plan"
X Link 2026-02-11T21:19Z [----] followers, [---] engagements
"i think the $200 Codex sub is pretty much unlimited. i'm trying my best to have multiple projects running in parallel with 2-3 /review chats running on every completed task for almost the whole day and i still can't deplete the usage 💀 my new workflow is coding with Cursor and then opening a terminal and running /review in 1-3 Codex chats as well as running Cursor's agent review in either quick or deep mode (depends on the task ofc). the goal is to minimise bugs before it's time for a human review https://twitter.com/i/web/status/2021734482898718864"
X Link 2026-02-11T23:53Z [----] followers, 14.8K engagements
"why i recommend running /review in [--] Codex chats instead of 1: chat 1: "I did not find any concrete regressions." chat 2: "P0 .""
X Link 2026-02-12T01:34Z [----] followers, [---] engagements
"@brianchew theyre both good though and we need opus for frontend"
X Link 2026-02-12T04:51Z [----] followers, [--] engagements
"@crowleylab do you top up with credits"
X Link 2026-02-12T04:52Z [----] followers, [---] engagements
"@embirico a Codex IDE that plays Sora videos while the agent is running 👀"
X Link 2026-02-12T06:00Z [----] followers, [---] engagements
"its composing time We've raised limits for Auto and Composer [---] for all individual plans. Individual plans now have 3x usage of Composer [---] versus Composer [--]. For a limited time (through February 16) we're increasing that to 6x. We've raised limits for Auto and Composer [---] for all individual plans. Individual plans now have 3x usage of Composer [---] versus Composer [--]. For a limited time (through February 16) we're increasing that to 6x"
X Link 2026-02-12T06:32Z [----] followers, [----] engagements
"@LLMJunky @Solaawodiya they also mentioned it comes out to be cheaper than codex since its included in cursors auto usage pool so you get more free usage when your plan hits the limit"
X Link 2026-02-12T08:07Z [----] followers, [--] engagements
"@DhravyaShah codex [---] is calling"
X Link 2026-02-12T08:37Z [----] followers, [--] engagements
"Cursor released public evals for Composer [---] and it seems to be between Opus [---] and Sonnet [---]. i'm finding it to be good for when you need super fast edits subagents or when you need to ask codebase questions. opus [---] for frontend. everything else codex 5.3"
X Link 2026-02-12T08:56Z [----] followers, [----] engagements
"@MacroMatt_ why are you finding codex [---] to be worse than sonnet"
X Link 2026-02-12T11:50Z [----] followers, [---] engagements
"@bil0090 facts [---] is insane"
X Link 2026-02-12T12:01Z [----] followers, [--] engagements
"@ryder_builds its better but opus [---] is still really good you can use either one for sure and they each have strengths"
X Link 2026-02-12T12:02Z [----] followers, [--] engagements
"Codex-5.3 is a beast at fixing GitHub CI issues. here's an example prompt: "check why the latest CI run failed fix it then monitor the run. repeat until green" Codex-5.3 doesn't take shortcuts so it won't delete parts of your CI just to make it green it'll keep going until it's actually fixed. https://twitter.com/i/web/status/2021948660762693902 https://twitter.com/i/web/status/2021948660762693902"
X Link 2026-02-12T14:05Z [----] followers, [----] engagements
"@DavidPrenticeJr the compaction in codex is really good whats server side compaction"
X Link 2026-02-12T14:51Z [----] followers, [---] engagements
"@badlogicgames i read the original essay and i was like how is anyone believing that just an AI generated creative writing piece"
X Link 2026-02-12T15:03Z [----] followers, [---] engagements
"@sama so curious what you guys are about to launch 💀 2x speed cerebras codex 🤣"
X Link 2026-02-12T16:33Z [----] followers, [---] engagements
"all of OpenAI is hype posting right now so excited to see what they're about to drop. also it seems to be Pro plan only. 2x speed Codex [---] (a man can dream) We have a special thing launching to Codex users on the Pro plan later today. It sparks joy for me. I think you are going to love it. We have a special thing launching to Codex users on the Pro plan later today. It sparks joy for me. I think you are going to love it"
X Link 2026-02-12T16:34Z [----] followers, [---] engagements
"@owengretzinger i think so i stopped looking at what it does during that loop but yea i remember it spamming sleep (at least in cursor i can't remember what the codex CLI does)"
X Link 2026-02-12T16:44Z [----] followers, [--] engagements
"@RhysSullivan this was occurring in codex [---] as well pretty annoying"
X Link 2026-02-12T17:23Z [----] followers, [---] engagements
"an ultra fast Codex model it's Christmas holy Introducing GPT-5.3-Codex-Spark our ultra-fast model purpose built for real-time coding. Were rolling it out as a research preview for ChatGPT Pro users in the Codex app Codex CLI and IDE extension. https://t.co/6knTmyQZ4N Introducing GPT-5.3-Codex-Spark our ultra-fast model purpose built for real-time coding. Were rolling it out as a research preview for ChatGPT Pro users in the Codex app Codex CLI and IDE extension. https://t.co/6knTmyQZ4N"
X Link 2026-02-12T19:14Z [----] followers, [---] engagements
"time to test Cursor's long-running mode to add [--] more puzzles to my puzzle site (which currently only has [--] puzzles) curious to see how it goes and how long this takes 👀"
X Link 2026-02-12T19:56Z [----] followers, [----] engagements
"@davis7 codex has some secret compacting magic it's insane"
X Link 2026-02-12T20:20Z [----] followers, [---] engagements
"@tejashaveridev @jediahkatz damn how big is this repo also do you find that you have to give it a way to verify its own work in order to get it to run this long trying it out now"
X Link 2026-02-12T20:47Z [----] followers, [--] engagements
"@tejashaveridev @jediahkatz thanks by acceptance criteria are you only referring to quantitative ones like tests or do you mean qualitative acceptance criteria too"
X Link 2026-02-12T20:54Z [----] followers, [--] engagements
"@jediahkatz @tejashaveridev thanks gonna try throw in a detailed prompt and [--] shot a product that i need built super excited. how much has this run cost out of curiosity @tejashaveridev"
X Link 2026-02-12T21:00Z [----] followers, [--] engagements
"this is still going for the past hour gonna try build an entire SaaS product next. i don't know how it's gonna go but i'm gonna be as detailed as possible with the prompt. i was loving ultra fast coding with composer [---] but this also feels super fun trying to maximise the # of hours for one run time to test Cursor's long-running mode to add [--] more puzzles to my puzzle site (which currently only has [--] puzzles) curious to see how it goes and how long this takes 👀 https://t.co/aDZyCrTGmw time to test Cursor's long-running mode to add [--] more puzzles to my puzzle site (which currently only"
X Link 2026-02-12T21:16Z [----] followers, [---] engagements
"@Solaawodiya i'm not sure if i'll use Spark if i'm doing fast edits i also want a fast way to review i.e. cursor's CMD+Y/CMD+N approvals"
X Link 2026-02-12T21:47Z [----] followers, [--] engagements
"see you in a few hours codex 5.3"
X Link 2026-02-12T22:02Z [----] followers, [--] engagements
"@Kannav02 they released it for ultra plans too"
X Link 2026-02-12T23:57Z [----] followers, [--] engagements
"@jarrodwatts that's crazy congrats 🫡 the fidget keychain looks interesting"
X Link 2026-02-13T00:05Z [----] followers, [---] engagements
"huge Cursor's got a new plan & usage tab in the settings to make usage more transparent. also seems like Auto + Composer have their own usage pool. loving this change 🫡 (although i wish the tooltip would come back and show a quick overview of total % auto + composer % and API %)"
X Link 2026-02-13T01:34Z [----] followers, [----] engagements
"@owengretzinger @itsnoahd one thing that's outdated is that Codex has plan mode built in now if you hit shift tab and you can enable subagents via /experimental and just ask codex to spawn some (e.g. to review an implementation)"
X Link 2026-02-13T03:16Z [----] followers, [--] engagements
"@jarrodwatts is it actually gonna attach to your keyboard or is it just a separate thing this whole merch kit looks sick"
X Link 2026-02-13T03:18Z [----] followers, [--] engagements
"4 long running Cursor agents going now time to get more running and cook up random ideas 👀"
X Link 2026-02-13T05:21Z [----] followers, [----] engagements
"@CristianXIV can always get a second account 👀 i'd burn it in [--] day if it was opus"
X Link 2026-02-13T05:46Z [----] followers, [--] engagements
"@anasibnanwar [--] of them are extending on my existing projects [--] of them are building entire SaaS products so i have no idea how that's gonna go. i hope i provided enough details 💀"
X Link 2026-02-13T06:31Z [----] followers, [--] engagements
"@cheatyyyy i found that bugbot pairs well with the codex reviews and it's unlimited"
X Link 2026-02-13T06:35Z [----] followers, [---] engagements
"i was getting codex-5.3 to replace something in the codebase and instead of doing it [--] by [--] it ran a codemod script to bulk replace them i love this model"
X Link 2026-02-13T08:56Z [----] followers, [----] engagements
"@LLMJunky oh damn how do you end up reviewing those (if you review them)"
X Link 2026-02-13T09:10Z [----] followers, [--] engagements
"running [--] of the cursor long running agents right now and there's no way any of this works (or will it 👀) [--] of them are extending on existing projects [--] of them are building completely new ones. longest one so far is at 11+ hours 💀 so excited to see the results"
X Link 2026-02-13T12:18Z [----] followers, [----] engagements
"@robinebers oh what it is"
X Link 2026-02-13T13:56Z [----] followers, [--] engagements
"my long running Cursor agent automatically set up a GitHub actions workflow so that it could run unit tests that it was adding"
X Link 2026-02-13T14:05Z [----] followers, [----] engagements
"@anasibnanwar yep"
X Link 2026-02-13T16:05Z [----] followers, [---] engagements
"i'm surprised how many people have --yolo off in Codex i nearly quit using [---] because it was so unusable. it's probably better with Codex [---] though and i'm assuming permission systems have gotten better within Codex now Coding agent users: do you run with --yolo (Codex) or --dangerously-skip-permissions (Claude Code) or equivalent Coding agent users: do you run with --yolo (Codex) or --dangerously-skip-permissions (Claude Code) or equivalent"
X Link 2026-02-13T16:10Z [----] followers, [---] engagements
"how are these running for so long WHAT is any of this code gonna run i mean if this actually works i'm gonna be mind blown but if it doesn't work my prompts probably could have been better to be honest"
X Link 2026-02-13T16:42Z [----] followers, [----] engagements
"@ernielm @rudrank nothing in claude because i just use opus in cursor if i need it"
X Link 2026-02-13T22:39Z [----] followers, [--] engagements
"honestly my workflow keeps changing but yea i use cursor for sync work/for opus/for asking about the codebase/for viewing files and i try have long running tasks in a separate terminal (ghostty) via codex using codex [---] more and more in cursor instead though which is more expensive but i like being able to use the cmd+Y and cmd+N approval flow"
X Link 2026-02-13T22:41Z [----] followers, [--] engagements
"@tejashaveridev oh yea its codex so you found opus to be better at these makes sense though codex thinks a lot"
X Link 2026-02-13T22:42Z [----] followers, [---] engagements
"@LLMJunky damn [---] files 💀 the subagent review using best practices skills looks interesting though"
X Link 2026-02-13T22:44Z [----] followers, [--] engagements
"@tejashaveridev ohhh yea that makes sense actually no wonder some of these are running for so long thanks i wonder if opus would use less $ since it finishes faster"
X Link 2026-02-13T22:48Z [----] followers, [--] engagements
"@ernielm yep the desktop apps are pretty nice i was sort of experimenting with using the codex app + cursor and its pretty good but i like the codex CLI more"
X Link 2026-02-13T23:08Z [----] followers, [--] engagements
"@ScriptedAlchemy you had the 1m line PR right"
X Link 2026-02-14T07:02Z [----] followers, [--] engagements
"real truth about AI in actual work outside of the X bubble everyone's talking about their teams like they were at the peak of efficiency and bottlenecked by ability to produce code here's what things actually look like - your org rarely has good ideas. ideas being expensive to implement was actually helping - majority of workers have everyone's talking about their teams like they were at the peak of efficiency and bottlenecked by ability to produce code here's what things actually look like - your org rarely has good ideas. ideas being expensive to implement was actually helping - majority of"
X Link 2026-02-14T07:45Z [----] followers, [---] engagements
"@realhasanshoaib yeah for sure its evident in the codex team shipping stuff really quickly"
X Link 2026-02-14T14:41Z [----] followers, [--] engagements
"@JuanRezzio gonna check them out monday 👀 i realised my prompts couldve been better since i noticed codex started adding useless tests in order to keep going"
X Link 2026-02-14T14:42Z [----] followers, [--] engagements
"holy [--] hours 💀💀 i think mine stopped at around [--]. are you using opus at all for the long running agents i had codex start adding useless tests at the end and thanks to @robinebers i learned it was due to the constant continue bumps that made it think it wasnt finished (bad prompting from me) https://twitter.com/i/web/status/2022684294670680360 https://twitter.com/i/web/status/2022684294670680360"
X Link 2026-02-14T14:48Z [----] followers, [---] engagements
"@nummanali interesting what does it usually come up with"
X Link 2026-02-14T14:51Z [----] followers, [--] engagements
"@udaysy hows the deployment process compared to vercel vercel is [--] clicks"
X Link 2026-02-15T06:40Z [----] followers, [--] engagements
"@matanSF i remember having some UX that didnt fit how i like working when i tried a few weeks ago do you mean codex good at frontend in your harness"
X Link 2026-02-15T06:45Z [----] followers, [---] engagements
"@udaysy yea makes sense that is cheaper but also vercel isnt [--] per project. im paying [--] and have like [--] projects hosted"
X Link 2026-02-15T11:15Z [----] followers, [--] engagements
"@theo loved this video and i'm experiencing the exact same problem at the moment. i don't know the solution but hopefully someone cooks one up"
X Link 2026-02-12T22:09Z [----] followers, [---] engagements
"do you need a free personal/portfolio website i'm going to be offering these for free - made with Cursor. they're not going to be mindblowing creative pieces of art but it should be a fun experiment. comment or DM me if you're interested in getting one i'll be accepting requests for a few days unless i somehow get way too many 🫡"
X Link 2026-02-13T21:05Z [----] followers, [----] engagements
"@vasuman if open source models are better than closed source would that tank the stock market"
X Link 2026-02-14T07:47Z [----] followers, [---] engagements
"whats your current tech stack for side projects im using: - Nextjs - Prisma - tRPC - Tailwind - Resend - Polar - Cloudflare R2 - Vercel for deployments - Neon for DB"
X Link 2026-02-14T09:01Z [----] followers, [----] engagements
"@ericzakariasson i like the auto assigning idea 🤔 might need to use that. we use yolomerge as the urgent PR merge in slack 💀"
X Link 2026-02-14T11:40Z [----] followers, [---] engagements
"@iambilalazhar yea and it doesnt seem to be that good at UI unless i point it at an exact component and ask it to copy it"
X Link 2026-02-14T14:44Z [----] followers, [---] engagements
"@LLMJunky @ProtonotariosA do you mean claudes better at writing react/nextjs code as well or you mean frontend"
X Link 2026-02-14T14:45Z [----] followers, [--] engagements
"@dobroslav_dev yeah i wonder why. i wonder if its because there isnt a ton of high quality react/nextjs code out there to train on 🤔 unironically they might need to get @aidenybai to write [--] million nextjs components manually to train Codex 6"
X Link 2026-02-14T14:50Z [----] followers, [---] engagements
"Auto plug: - Newsletter with AI tips/tricks and my journey shipping code for work and side projects: - Subscribe to my YouTube just started this year: https://www.youtube.com/@kr0deryt https://kr0der.com/newsletter https://www.youtube.com/@kr0deryt https://kr0der.com/newsletter"
X Link 2026-02-14T16:20Z [----] followers, [---] engagements
"i still cant get over how good Codex and Opus are. my puzzle site wouldnt even exist right now because i have no idea how to code puzzle generation and verification logic but i built the entire thing with those [--] models. now i have a site that i get to play daily and have fun and my family and gf also play it too which is super nice :) i love AI 🫡 https://twitter.com/i/web/status/2022763922067947858 https://twitter.com/i/web/status/2022763922067947858"
X Link 2026-02-14T20:04Z [----] followers, [----] engagements
"@SharpCoder you can get it to write it to the codebase and then reference that but i found codex is super good at compacting and long running tasks so it feels fine tbh"
X Link 2026-02-15T06:49Z [----] followers, [---] engagements
"makes sense i dont use vercel for free projects for the same reason (if it blows up its doomed) but if a paid project blows up then its usually a good thing. ive heard good things about tanstack start + convex though so ive been meaning to try that out too. so is your main stack go backend + ts frontend"
X Link 2026-02-15T09:03Z [----] followers, [--] engagements
"if someone posts that AI has killed XYZ or USE AI for XYZ OR YOURE NGMI check their bio and 99% of the time theyre selling an AI tool that doesnt exactly what the post says always take opinions on X with a grain of salt - verify them yourself too"
X Link 2026-02-15T20:04Z [----] followers, [---] engagements
"@ernielm @OpenAIDevs @cursor_ai u sorta can by using the IDE extension or in a terminal"
X Link 2026-02-16T01:03Z [----] followers, [--] engagements
"Codex [---] is so good man wtf copy this AGENTS md line from @steipete to get it to add regression tests every time you get it to fix a bug: "Bugs: add regression test when it fits." i added this and it's 10x - Codex fixes a bug and automatically adds tests right after"
X Link 2026-02-09T23:53Z [----] followers, 125.4K engagements
"get Codex to review its own work with this prompt: "when done review your changes with [--] subagents fix any issues then repeat until no issues found" i always use [--] subagents rather than [--] because LLMs aren't deterministic so i like having [--] reviewers to catch more bugs"
X Link 2026-02-11T01:34Z [----] followers, 20.9K engagements
"i love Codex but i just cant trust it to ever touch frontend i just instantly swap to Opus whenever frontend needs to be done. also i feel like i notice both models perform crimes with React/Nextjs and i dont even know hooks as well as some people so i think you either need a really good AGENTS md and/or skills has anyone solved this https://twitter.com/i/web/status/2022640105442152650 https://twitter.com/i/web/status/2022640105442152650"
X Link 2026-02-14T11:52Z [----] followers, 10.9K engagements
"save these Codex tips: [--]. use yolo mode but disclaimer - be careful of getting prompt injected if youre installing a dodgy skill or getting it to navigate the web. the reason for yolo mode is that codex is beast at long tasks so i dont like when it gets blocked. [--]. use plan mode - codexs plan mode is actually pretty good and i enjoy using it but its not necessary if you dont like it. you can also just chat to the model which worked fine before plan mode existed. [--]. talk to Codex as if its really really dumb. this sounds weird because its super smart but if youre even a bit vague it might go"
X Link 2026-02-14T22:54Z [----] followers, [----] engagements
"@theo slack activity tab has been bugged for at least [--] months"
X Link 2026-02-15T01:22Z [----] followers, [---] engagements
"people in the year [----] are gonna be like what religion do you follow codex claude grok deepeek glm minimax or gemini"
X Link 2026-02-15T15:55Z [----] followers, [---] engagements
"@robinebers it came - Kimi Claw"
X Link 2026-02-15T19:06Z [----] followers, [--] engagements
"the reason senior engineers benefit from AI agents more is because they can give more detailed instructions when prompting. im not the most senior engineer so when a task goes wrong its almost always a prompt issue due to lack of experience or knowledge in a certain field. when do prompts go badly the most often for you https://twitter.com/i/web/status/2023169094971765040 https://twitter.com/i/web/status/2023169094971765040"
X Link 2026-02-15T22:54Z [----] followers, [---] engagements
"@alex__bit yea it did also not sure what it used sorry didnt read it"
X Link 2026-02-16T01:05Z [----] followers, [--] engagements
"@vasuman but the twitter bubble told me that IDE = ngmi 🤣"
X Link 2026-02-16T02:32Z [----] followers, [--] engagements
"@amritwt did you end up running codex at the anthropic hackathon"
X Link 2026-02-16T02:52Z [----] followers, [---] engagements
"@bailout00 thanks didn't know that still good to know since some people instantly swap mid-chat"
X Link 2026-02-16T05:31Z [----] followers, [--] engagements
"gonna test replacing my Codex /review chats with just running agent review in "Deep" mode in Cursor. reason being that the Codex /review sometimes feels like it overthinks and ends up slowing me down by pointing out unnecessary things and i think Bugbot has really good accuracy. this is basically Bugbot in the IDE so i'll test running only this for reviews instead and see how it goes. last time i asked this deep mode costs around $1 per run which is easily worth it. @aye_aye_kaplan any chance we can run [--] deep reviews in parallel 👀 https://twitter.com/i/web/status/2023321868577525911"
X Link 2026-02-16T09:01Z [----] followers, [----] engagements
"in the Codex terminal and app you can hit shift tab to enter plan mode. it has better UX than Claude Code because you can hit tab to add any extra notes to your answer which is MASSIVE imo i always had to choose "None of the above" in CC because i needed to add extra notes. obviously CC pioneered plan mode so i love CC as well if Codex terminal (or app) ever gets Plan mode I will 100% be switching from Claude Code if Codex terminal (or app) ever gets Plan mode I will 100% be switching from Claude Code"
X Link 2026-02-17T01:17Z [----] followers, [----] engagements
Limited data mode. Full metrics available with subscription: lunarcrush.com/pricing