Reading up on Codex

Now in beta: composer predictions in Codex for Pro users. Codex can now suggest your next message based on your conversation and how you talk to it. One of the most loved new features we've ever tested internally.

@OpenAIDevs

If your company or startup is running engineering with everyone using their own license of Codex/CC then you're going to be completely left behind by fast moving teams who are using cloud-based agents. Local dev is like banging rocks together and grunting versus being on a rocket. Seriously, just pay the money for something like @DevinAI, @AmpCode, @FactoryAI, etc. Do NOT try to build your own "code factory". Props to @DaleSpoonemore for taking @HelloUntangle to the next level on this. Fun to watch.

@ryancarson

It was hilarious at first, funny at second but it’s starting to get tiring and borderline annoying, Grok Bot and Codex/Dots not stopping at throwing punches at one another

@GergelyOrosz

I’ve tried rebuilding my website about 5 times since coding models came out. Design always was bad and nothing felt as simple as my old setup. Finally Codex built it, along with a small tool to add my writing and deploy the static site to S3. so enjoy https://karrisaarinen.com

@karrisaarinen

You can finally close your laptop now and your agents will keep working. Codex cloud environments are here. Reusable environments mean less setup and faster starts, with your repo, dependencies, scripts, and settings already in place.

@OpenAIDevs

Codex voice for road trips:

@gdb

holy hell... chat mode for codex is 🥵 -- I just talk to it...

@kevinrose

Yet to see an example of something grok bot, instinct, muse etc can do that codex or claude can’t do with the same prompt

@frantzfries

Codex pro tip: You can reference one thread from another just by @-mentioning or drag/dropping! You can also ask it to message or monitor others. Combine this with schedules and you can manage whole fleets of threads working together.

@btraut

How do I work in parallel with Claude/Codex on my Mac Mini without worktrees? Right now using a worktree skill to create them and then clean them up afterwards but many smart people are saying they suck.

@iannuttall

If you use Codex, @thsottiaux needs no introduction. We talked 100% about Codex (fun fact: he started building it!), and 0% about resets. Timestamps: 00:00 Intro 07:21 Working at Google 12:41 What drew Tibo to OpenAI 15:19 The early days of Codex 18:20 Why Codex was built in Rust 21:15 Why Codex is open source 25:50 Codex plays nice with other models: why? 32:09 How the harness works 36:44 Harness and model improvements 41:19 The SDLC behind Codex 46:39 Code reviews at Codex 52:09 Maintenance and architecture 56:43 How AI tools expand what engineers can do 1:02:30 The Merge: ChatGPT + Codex 1:07:16 How Tibo uses Codex and ChatGPT 1:10:44 Advice for engineers who want to work in AI Brought to you by: • @turbopuffer – a vector and full-text search engine built on object storage. It’s fast, cheap, and extremely scalable.

@GergelyOrosz

This is amazing. Copy this screenshot into Codex. It can solve most issues with Astra not following instructions. From @pvncher.

@daniel_mac8

[Channels Mariah]: It's tiiiiiiiiiime! Today @every is releasing Compound Writing, so now you - yes, you - can install my tendency to overthink a paragraph. I started this past winter by pulling down the Compound Engineering plugin and telling Claude (4.5 at the time, what a model that was, “I want to build this but for writing.” I’ve spent the last six months building and writing with and adding to and tweaking it. The plugin gives Claude Code and Codex skills for interviewing you about an idea, shaping a draft, and questioning whether your argument holds up. When you figure out something useful about your writing, you can save it as guidance for the next piece instead of having the same conversation with the machine indefinitely. I’m standing on the shoulders of giants here in the form of @kieranklaassen and @danshipper, whose work on Compound Engineering and agent-native architectures informed the evolution of this project - what, why, and how. My process, and this plugin, is one way to write with AI. Someone else could start with the same tools and build something completely different, and I’d love to see it. You can use the whole process, borrow one editing skill, or write every word yourself and ask for feedback. No obligation to adopt my entire collection of hang-ups. My hope is that Compound Writing gives you a new playground to discover your own process and grow as a writer. I’ve written a guide to getting started:

@kplikethebird

Found a GitHub repo that gives your AI agent 50 marketing specialists for free. It's called marketingskills, built by Corey Haines, and it sits at 46,800 stars with 7,300 forks. The idea is simple. Skills are markdown files. Each one teaches an AI agent how to do one marketing job, with the frameworks and checklists a good marketer carries in their head. There are 50 of them, covering copywriting, CRO, cold email, pricing, SEO audits, A/B testing, churn prevention, ad creative, launch planning, referral programs, and the list keeps going. They're also wired together. One skill called product-marketing holds the context about your product, your audience, and your positioning. Every other skill reads that file first before doing anything. So the copywriting skill and the pricing skill work from the same understanding of what you sell. A few of them caught my eye. marketing-council spins up a simulated board of advisors so you get several expert takes on one question instead of a single answer. marketing-loops sets up recurring workflows an agent runs on a schedule without you re-prompting it. marketing-ideas is a bank of 139 ideas for SaaS products you can pull from when you're stuck. Installing is one command. It works with Claude Code, Codex, Cursor, Windsurf, and anything else that follows the Agent Skills spec. You type "help me optimize this landing page for conversions" and the right skill kicks in on its own. Is a folder of markdown files worth more than a $5,000 a month marketing hire?

@alex_verem

life update: after 4 incredible years at Vercel leading the Next.js and React teams, I’m joining OpenAI to work on Codex as an IC. ChatGPT and Codex have completely changed how I build. Getting the chance to help shape where this goes next was too exciting to pass up. I learned an absurd amount at Vercel, and I’m excited to bring some of it with me. onwards :)

@feedthejim

Codex has changed my life as a solopreneur

@jonnym1ller

I wanna share all my startup experience, one topic at a time, Today it's ---SEO--- In the past years, I generated over 1B impressions and 1M clicks from SEO. Everything I've done: 📁 Domain names ↳ Use: .com, .org, .dev, .ai, .io ↳ Use keyword-like naming, e.g. osssoftware dot org ↳ Max 14 letters in a domain name ↳ No “-” or “_” ↳ One topic per domain when possible ↳ Pick www or non-www and stick to that one version (one redirects to another one) 📁 Tech SEO ↳ Make sure the page renders HTML when bots make requests ↳ One way to check it:  ↳ Drop your URL into Claude  ↳ Ask if it’s doing everything right for tech SEO ↳ There are many more details, but really, Claude or Codex will lay it all down for you if you show it your website ↳ HTTPS on ↳ Mobile-friendly ↳ Fast enough that it doesn’t feel broken  ↳ PageSpeed Insights is enough  ↳ Open GSC web-vitals and if it's green then all good ↳ Short, stable URLs  ↳ Url slug should match the meta title and h1  ↳ No random strings ↳ One H1 per page ↳ Unique title + unique meta per page ↳ Image alt text on anything that matters ↳ Canonical URL on every indexable page that has no query string in it ↳ Fix 404s and redirect chains ↳ No orphan pages  ↳ Every important URL should have at least one internal link pointing to it (seogets shows all such pages) 📁 Indexing ↳ Add sitemap.xml to Google Search Console ↳ See if all the pages are indexed  ↳ If not, run @indexrusher ↳ Do the same for Bing and Yandex via indexrusher ↳ robots.txt should be there and allow your public pages, again, ask chatgpt to check your robots.txt ↳ Request indexing in GSC for new pages or run indexrusher to automate this ↳ Check Coverage / Pages reports  ↳ Fix “crawled – currently not indexed” on pages that should rank  ↳ Don’t waste crawl budget (don't create too many pages, too many folders, tags, groupss...at least right away, you better do it later once the first bulk is indexed) 📁 Analytics ↳ Google Search Console ↳ Ahrefs small plan ↳ @seogets + @DataFast_ ↳ Once a week, look at:  ↳ Queries with impressions but few clicks (take the queries pointing to these pages and embed them into h1 and meta title)  ↳ Pages that used to get clicks and then decayed (update the content here) 📁 Low hanging fruit ↳ Pages already in positions 8–20  ↳ Those are closer to page 1 than any new article  ↳ Add the exact query(if makes sense) to H1 / title / H2  ↳ Add one missing section the current top results have ↳ Queries with lots of impressions and almost no clicks  ↳ Rewrite title + meta, keep the keyword ↳ Old pages that already get a little traffic  ↳ Update them before writing a new one (google kinda loves to see pages being regularly updated) ↳ Pages Google knows but didn’t index  ↳ Request indexing manually page by page, or run @indexrusher (if it doesn't help, remove them, cuz google wont index them ever) ↳ Thin pages fighting each other for the same query  ↳ Keep the stronger URL  ↳ Merge or 301 the weaker one 📁 Quick 5min SEO hacks ↳ Paste a URL into GSC → URL Inspection → Request indexing ↳ Put the target keyword in the first 60 characters of the title ↳ Add the main query into the H1 if it’s missing ↳ Add 2 internal links from old posts to the page I want to push ↳ Add alt text to the first image on the page ↳ Add a 3-question FAQ at the bottom using GSC queries ↳ Submit / resubmit sitemap.xml in GSC ↳ Search site:mydomain and see if the pages are indexed and how they look like ↳ Google my brand + product keyword  ↳ If a page mentions me with no link, ask them to add one ↳ Ask ChatGPT “best tool for [keyword]”  ↳ Open the cited URLs  ↳ Email those authors the same day (offer them either money or a link exchange) ↳ Post the article URL on X / LinkedIn / Reddit with the keyword in the first line ↳ Drop the URL into Claude: “what 5 on-page fixes can I do in 5 minutes?” 📁 Blog ↳ Write 1 good blog post a week  ↳ Do it manually with some AI assistance  ↳ Focus on the most promising keywords where you already rank high ↳ Use @seobotai to generate more blog articles  ↳ See if some of them get traction  ↳ Manually turn the promising ones from AI-generated into high-quality human-made articles ↳ Before writing, search Google for the keyword  ↳ Match the format that already ranks  ↳ Answer the question better than the current top 5 ↳ Put original stuff in the article  ↳ Screenshots  ↳ My own numbers  ↳ What I actually tried (basically ensure google that this article comes from a human's own experience) ↳ Refresh old posts that already get impressions  ↳ Update the date only if the content actually changed (avoid hacks here, do real updates) ↳ Useful page types on top of the weekly post:  ↳ “best X”  ↳ “X vs Y”  ↳ “X alternatives”  ↳ “how to” for the exact job my product does 📁 Backlinks ↳ Ask ChatGPT: “What’s the best tool for...”  ↳ See the URLs it used to form the response  ↳ Reach out to all those authors  ↳ Ask them to include my URL in exchange for money or a backlink from you ↳ Do the same in Claude, Perplexity, and Gemini and Grok ↳ Search Google for your keywords  ↳ Find the pages that already rank  ↳ Do the same as above ↳ List on as many directories as possible using @listingbott ↳ Link all my pages across all my websites when possible ↳ Do internal linking on each site too  ↳ Between blog articles  ↳ Between pages ↳ Extra places that still work:  ↳ Guest posts on sites that already rank for my keywords  ↳ “best tools” roundups  ↳ Unlinked brand mentions   ↳ Search Google for my brand name   ↳ Ask them to turn the mention into a link ↳ When I link out or guest post, I want the page to have:  ↳ Real traffic  ↳ A topic close to mine  ↳ A link that looks natural on the page (avoid spam at all cost) 📁 Social media SEO ↳ Social media gets indexed instantly  ↳ Reddit  ↳ X  ↳ Substack  ↳ LinkedIn ↳ Monitor hot, growing keywords and trends  ↳ Google Trends  ↳ Radar on X ↳ Produce posts around those trends  ↳ Place some of the product links inside ↳ Find relevant social media posts via Google Search  ↳ Plug my product into the comments when it brings some value  ↳ No spam ↳ Put the keyword in the first line of the post ↳ Turn each good blog post into (and link back to the original url with a backlinks, cuz social media backlinks in my experience have really high weight for google):  ↳ 1 X thread  ↳ 1 LinkedIn post  ↳ 1 Reddit answer where it actually helps ↳ Reply under bigger posts in my niche  ↳ Useful reply first  ↳ Link only if it adds something 📁 AI search ↳ Same game as SEO ↳ Ask ChatGPT / Claude / Perplexity / Gemini:  ↳ “What’s the best tool for [my keyword]?”  ↳ “Best alternatives to [competitor]”  ↳ See if they mention me  ↳ If not, look at who they mentioned and get onto those same pages by paying them or doing a link exchange --- That's it for now. I'll share the next topic in my next tweet. What should it be? Also, i put it all in great details into my Founder Guide:

@johnrush

I don't think most of us outside of OpenAI realize just how Codex-pilled everyone inside OpenAI is Because the Codex that we see and use is nowhere nearly as powerful as the one they see and use inside (Theirs integrated anywhere & everywhere inside OpenAI, nearly full access)

@GergelyOrosz

Semrush charges $139.95/month. Ahrefs charges $129/month. A guy named Ben built the open-source alternative that you can run on your computer. It is called OpenSEO. Enter a competitor's website and it reveals the keywords bringing them traffic, the pages winning those searches, and the backlinks helping them rank. And enter your own site and it tracks rankings, crawls technical problems, reads Search Console, and shows whether ChatGPT or Google AI Overviews mention your brand. Then connect Claude or Codex. Your agent can pull the real numbers, compare opportunities, cluster keywords, and save a strategy back into the dashboard instead of guessing from a prompt. The hosted version starts at $10 a month. The entire codebase is MIT licensed, so you can also self-host it and pay DataForSEO directly for only the data you use. No, it does not magically reproduce every database and advanced feature Ahrefs built over a decade. But it breaks the most important rule of the old SEO market: You no longer need their permission to own the tool.

@ihteshamali

codex' visualization feature got really good.

@steipete

We’re excited to announce the OpenAI Build Week winners 🥁 Meet the builders behind the eight winning projects and see what they shipped with Codex. https://openai.com/build-week

@OpenAIDevs

incredible use of computer history “based on my computer history, what single-use software would make my life easier?” computer history turns your activity across apps and websites into memories and a timeline, so codex can spot repeated workflows and build the little tools that make them easier

@reach_vb

My full interview with Tibo (@thsottiaux) 0:45 Tibo's Lessons from Google DeepMind 4:22 Building OpenAI’s Relentless Culture 7:23 Astra & Next Gen Models 11:18 How Fast AI Changes Developer Workflows 14:27 ChatGPT & Codex Merging 20:25 OpenAI vs. Anthropic 23:37 Why OpenAI Keeps Resetting Limits 30:25 Recursive Self-Improvement 32:00 Dangers That Caused "The Pause" 34:13 Will Ultra Fast Become the Default? 43:20 Why Everyone Needs to Try AI

@MatthewBerman

It's me again. I come bearing great news. First of all, we have hit 20M active users for Codex some time this week. Second of all, this is cause for celebration and during the day we will credit every Codex and ChatGPT Work user with a BANKED reset that you can use at your own leisure. And we will have some other good news later too! Now, on usage limits draining faster, while we're not seeing anything abnormal, we do take it incredibly seriously and there is an ongoing investigation. I will share if we do find anything and my below post is really a clarification on a specific pattern that we did see that I wanted to call out. Go do something amazing today.

@thsottiaux

Everyday conversations just got easier with the new Apple Messages plugin. Search messages, catch up on conversations, draft and send replies—all with ChatGPT on your Mac. Now available in ChatGPT Work and Codex on desktop.

@ChatGPT

What is an obvious thing that we should do with Codex, API or our models that we should just do but haven't yet? What is 100% within reach, but we just seem to be missing?

@thsottiaux

Good moment today to remind everyone to stay frosty with AI. I was working on making a server stoppable via the CLI, and Codex decided the best path forward was to implement an unauthenticated network API to every server to shut itself down. Excellent work. I responded with "I don't know about that." (I usually give more direct feedback but this was so absurd) and it thought for like 20 seconds and responded "Server stopping should not be part of an unauthenticated network API." Yes. lol. By the way, the result I was looking for was that we automatically start a locally discovery Unix-socket bound server so I wanted it to discover that (using prior mechanisms already made) and gracefully stop that. The driver is absolutely accountable here (me). My lazy prompting led AI to think I wanted to be able to stop ANY configured server I was talking to. Shame on me for thinking it was obvious here. But, I also reviewed it and found this idiocy, so, that's where good AI drivers come in. Also me.

@mitchellh

Brent is really pushing the edge of video editing with codex! Give him a follow

@pvncher

The question I get most right now is "what should I turn into an AI agent?" Here is the 5 part test I use (works for Claude/Codex/Grokbot etc): 1. It has a repeated trigger, so the same kind of task keeps happening over and over. 2. The inputs are stable, so the information comes in a predictable shape every time. 3. The tools are clear, so there is a defined set of things it can actually go do. 4. There is a measurable finish line, so you can tell when it is done and whether it worked. 5. There is judgment in the middle, so each run is a little different and needs a real decision made in the moment. The first 4 are really just asking "can this be automated at all?" The 5th is the one that matters, because judgment in the middle is what makes it an agent instead of a simple automation. Once you see it this way, you start spotting agent work pretty much everywhere.

@gregisenberg

Hey @steipete you know I’m the ultimate clawboi - if you want my true competitive analysis openclaw ++ vibes / personality + local device (but tbh other than being cute doesn’t matter w cloud APIs, but we love a Mac mini moment) + hackable + multiplayer use (good in a group chat for slack) + crons - connector / MCP / gog constant death loop and management - maintenance while traveling grokbot @laurenleeplace ++ connector experience + consumer friendliness (AI pilled my very not ai native friend) + bot to bot hand offs - no multiplayer - vibes - opaque - not hackable eve @shardara @cramforce ++ enterprise connectors + easy to collaborate on building the bot + simple framework for config + evals, monitoring etc + I love chatSDK + tool control / approval UI is great - not consumer friendly - not really self service for non eng - vibes but that’s kind of a me problem codex @ajambrosino ++ just the best ++ code code code ++ computer / browser use + vibes are fine - actually hate the connector experience - no multiplayer - no little computer in the cloud hermes - vibes off just can’t do it

@clairevo

Codex and ChatGPT can now understand the context of your recent work. Opt into Computer History to give ChatGPT richer context, so it can pick up where you left off, understand patterns in your work, and suggest skills or scheduled tasks for work you repeat.

@OpenAIDevs

I previously promised a reset for every 1M in additional active users for Codex, until 10M. We blew past that and have been silent since 10M. Little surprise for you tomorrow.

@thsottiaux

Why did you switch to Codex? What do you like about it and what could we improve? Don’t say reset.

@thsottiaux

New native Codex app for Linux looking great on Omarchy. Good job, @OpenAI! We'll have it hosted on the OPR and available for immediate install in Quattro.

@dhh

Spotify has run this exact play before. In March 2020 they open sourced Backstage, the internal portal their engineers built to manage 14,000 software components. It became the industry standard for developer portals. Over 3,400 companies adopted it, including Netflix, American Airlines, and Expedia. Then Spotify built a paid enterprise product, Portal, on top of the free standard they created. Xirp is the sequel, aimed at a bigger prize. Look at what it actually does. Every agent session runs in its own git worktree, so 50+ agents can work on the same codebase in parallel without touching each other. And context lives outside the agent itself, so you can switch from Claude Code to Codex to Gemini mid-project and the full working state carries over. That second part is the whole strategy. Anthropic, OpenAI, and Google are spending billions to make their agent the one your team depends on. Xirp makes them interchangeable. Your dependency moves up a layer, to the environment holding your sessions, your service catalog, and your architectural decisions. Which happens to plug into Portal, the product Spotify sells. The model vendors are fighting to own the agent. Spotify is betting the agent becomes the commodity and the context becomes the product. Last time they made that bet, 3,400 companies took the free version and Spotify sold them the enterprise one. The music streamer from Stockholm is quietly building one of the better dev tools businesses in the industry, and almost nobody noticed the first time.

@aakashgupta

A great way to learn design: Give Codex a well-designed website, ask it to analyze what makes the design great, then ask it to take a complete screenshot of the website & add annotations on the image that breaks down why the design works Learning from examples is always better than learning from theory! And using the annotation method means you don't have to constantly switch between reading the analysis and viewing the artifact

@zarazhangrui

Codex team: If I am in Codex tab and say "schedule this task in the cloud" (it's a task that just uses plugins) is it actually going to schedule it in the cloud or I have to switch to ChatGPT Work to do this? It doesn't seem to work?

@petergyang

Built this tiny app in few hours with Codex, what a time

@tar_uniqueee

Of all the amazing things that AI is enabling, rigging codex to be my personal assistant has become one of the most impactful, examples: 1. returns/warranties 2. selling stuff on fb 3. trip planning 4. shopping 5. real estate prospecting 6. personal finance harness details below:

@BrandonMChu

ok y'all know i love an anti-todo list (stuff AI can take off my plate FOREVER) in this ep of how i ai, @nickbaumann_ shows me all his expert voice + sites + chatgpt work tricks so i can hand even more off to codex jarvis-style voice assistants, lazy prompting, video use cases, yolo computer use -- this ep hits all my faves A huge thanks to our wonderful sponsors: ⚡️ @boltdotnew - turn your idea into a real product:

@clairevo

i've used so many harnesses over the past few months, i'm having trouble remembering *where* i had past conversations. claude desktop *chat*, claude desktop *cowork*, claude desktop *code*, claude code TUI, codex, chatgpt, hermes, openclaw, conductor 😭

@Shpigford

Setting up Fable to use Codex 5.6 Sol as a subagent was the best idea ever—Fable is super great at supervising but pretty slow at executing on its own and Codex is really fast but needs a nanny. Perfect combo.

@SnazzyLabs