From b8da6040102430c95560a2a57ed797233f82a6cc Mon Sep 17 00:00:00 2001 From: Aishwari Pahwa Date: Tue, 19 May 2026 17:10:42 +0530 Subject: [PATCH 1/8] latest blog --- .optimize-cache.json | 1 + .../+page.markdoc | 103 ++++++++++++++++++ .../cover.avif | Bin 0 -> 6391 bytes 3 files changed, 104 insertions(+) create mode 100644 src/routes/blog/post/openai-just-shipped-codex-to-the-chatgpt-mobile-app/+page.markdoc create mode 100644 static/images/blog/openai-just-shipped-codex-to-the-chatgpt-mobile-app/cover.avif diff --git a/.optimize-cache.json b/.optimize-cache.json index f8b0dcae5d..bf19c09348 100644 --- a/.optimize-cache.json +++ b/.optimize-cache.json @@ -872,6 +872,7 @@ "static/images/blog/open-source-backend-vs-managed-saas/cover.png": "5046fc4e3f3fa605c22d00ecb4b5156734c7d94000b49cae66a9e3bc49f81984", "static/images/blog/open-source-firebase-alternative/cover.png": "e0b98679795c00fd6d2d304b17273eaa6847bb1fc5706efa6cc3f3040ec6b4bb", "static/images/blog/open-source-regulated-environments/cover.png": "bbb57f0f8dc70f31923f34b5007f089a8539bdbca193c6f3a2594502bca551a2", + "static/images/blog/openai-just-shipped-codex-to-the-chatgpt-mobile-app/cover.png": "32abc0a488d7b7b8f2a4f5661801f415bb1cc1dc148768159bead6e5815c4bde", "static/images/blog/openjs.png": "4129893c1ea043b82af755a462fda8d78a7a5199638f84da549b2a2b29fbe976", "static/images/blog/oss-netlify-cover.png": "166487672a542d051b029ba285b703fc7f6c49d2fa27407f22c8f1848725120f", "static/images/blog/OSS-program.png": "b2cb73db872591b279e9b7ce87eb1c281732f394ef96a5d4e941ba3892924be7", diff --git a/src/routes/blog/post/openai-just-shipped-codex-to-the-chatgpt-mobile-app/+page.markdoc b/src/routes/blog/post/openai-just-shipped-codex-to-the-chatgpt-mobile-app/+page.markdoc new file mode 100644 index 0000000000..782cbd2220 --- /dev/null +++ b/src/routes/blog/post/openai-just-shipped-codex-to-the-chatgpt-mobile-app/+page.markdoc @@ -0,0 +1,103 @@ +--- +layout: post +title: OpenAI just shipped Codex to the ChatGPT mobile app +description: OpenAI just brought Codex to mobile. Here’s why async AI coding workflows are about to change how developers build. +date: 2026-05-19 +cover: /images/blog/openai-just-shipped-codex-to-the-chatgpt-mobile-app/cover.avif +timeToRead: 5 +author: aishwari +category: ai +featured: false +faqs: + - question: Is Codex on mobile free? + answer: Codex in the ChatGPT mobile app is rolling out in preview on iOS and Android across all plans, including Free and Go, in all supported regions. Update the ChatGPT mobile app and the Codex app on macOS to try it. + - question: Does the phone replace my laptop for Codex work? + answer: No. The phone is for steering, approvals, clarifications, and kickoffs. The moments the agent needs you while it is running. Heavy review and architectural work still belong at a desk. + - question: What is Remote SSH for Codex? + answer: "It is Codex's ability to connect into remote development environments through your SSH config. The desktop app auto detects hosts and lets you create projects and run threads inside those remote machines as if you were working locally." + - question: Are programmatic access tokens and Hooks free? + answer: Programmatic access tokens are available on Enterprise and Business plans. Hooks are available on all plans. HIPAA compliant use is supported for eligible ChatGPT Enterprise workspaces only when Codex is used in local environments. +--- +OpenAI brought Codex to the ChatGPT mobile app, opened up Remote SSH so it can run inside managed dev environments, and shipped Hooks and programmatic access tokens for teams. The headline is mobile. The real shift is that agentic coding is now an async, anywhere workflow. + +On May 14, 2026, OpenAI [announced](https://openai.com/index/work-with-codex-from-anywhere/) that Codex is now available in preview inside the ChatGPT mobile app, alongside a batch of enterprise focused updates. The headline grabs attention, since you can now dispatch Codex tasks from your phone. The more interesting read is what the release tells you about how OpenAI thinks Codex should be used. + +If you have been running Codex as "a chat box that writes code," the second half of this announcement is the part that should change how you work. + +# What actually changed for Codex + +Five things shipped, and the boundaries between them matter. + +**Mobile access is now in preview on iOS and Android** through the ChatGPT app, across all plans including Free and Go. Before this release, Codex workflows were still primarily tied to desktop and IDE environments. + +**Remote SSH is generally available.** The desktop app auto detects hosts from your SSH config and lets you create projects and run threads inside remote machines just like you would locally. Before this, remote environment workflows existed, but they were far less integrated and not officially streamlined inside the Codex app. + +**Programmatic access tokens** provide scoped credentials issued directly from ChatGPT workspace settings, so Codex can be called from CI pipelines, release workflows, and internal automations without a human session. + +**Hooks are generally available** and can be used to scan prompts for secrets, run validators, log conversations, create memories, or customize Codex behavior for specific repositories and directories. + +**HIPAA-compliant local use** is supported for eligible ChatGPT Enterprise workspaces when Codex is used in local environments across the CLI, IDE, and App. + +A few practical reads: + +* **Mobile Codex is the async story, not the laptop replacement.** You are not going to refactor a service from a 6-inch screen. You are going to kick off a task while waiting for coffee, approve a diff on the train, and answer Codex's clarifying question before it stalls. The phone unlocks the gaps in your day, not the deep work. +* **Remote SSH being GA changes who can adopt Codex.** Teams that develop inside managed remote environments, devboxes, jump hosts, and locked down VMs did not have a streamlined first party Codex workflow before. Now the connection is part of the desktop app. +* **Programmatic access tokens move Codex closer to becoming part of CI/CD and internal automation workflows.** Scoped credentials issued from ChatGPT workspace settings mean Codex can run inside release workflows and internal automations without a human in the loop. +* **Hooks are the customization surface.** This is the layer most teams will spend time configuring, especially for secret scanning and policy enforcement. +* **HIPAA compliant local use is a healthcare unlock.** Narrow but real. Enterprise workspaces can now run Codex locally for patient facing and operational workflows. + +# Why a mobile Codex matters more than it sounds + +It is easy to be cynical about "AI in a phone app." Most of them are wrappers. This one is not, and the reason is the rhythm of agentic work. + +When Codex was synchronous, you typed, it wrote, you read, you committed. Being tied to a laptop was fine. The model finished before you got up from your chair. As agents take on longer tasks, that loop breaks down. A 20 minute refactor or a multi file investigation does not finish in a coffee break. It runs while you do other things. And while it runs, it needs you for three things: + +1. A clarifying answer when context is missing. +2. A direction call when there are two viable approaches. +3. A review of the result before it gets merged. + +None of those need a keyboard. All of them block the agent if you cannot respond. That is the gap mobile Codex fills. + +OpenAI says more than 4 million people use Codex every week, and the framing in their release is telling: "a quick check in can keep a thread moving, prevent unnecessary rework, or help Codex make progress with the right context." That is the language of a team that has watched users get stuck waiting on themselves. + +The three patterns OpenAI calls out are worth taking seriously: + +* **Start investigating a bug while waiting for your coffee.** Codex inspects the relevant files, reproduces the issue, runs tests, and starts working toward a fix. You approve clarifications from your phone. By the time you sit down, there is a diff to review. +* **Reach a decision point during your commute.** Codex hits a fork, surfaces two viable approaches, and asks for direction. You pick one. The task keeps moving. +* **Brief Codex before a customer call.** Codex can synthesize context from the environments and workflows it has access to. You walk in prepared. + +The common pattern is that the agent never has to wait for you to be at your desk. The next bottleneck becomes infrastructure and integration work, not coding speed. + +# Run Codex in enterprise environments + +This is the part of the announcement that matters most for enterprise adoption. Remote SSH going generally available is what makes Codex viable for teams that cannot install random tooling on developer laptops. + +The flow is straightforward. The desktop app reads your SSH config, lists the hosts it finds, and lets you create projects and run threads inside those remote machines. Once connected, those environments become accessible across your authorized ChatGPT devices through OpenAI's secure relay. You can start work on your desktop, steer execution from your phone, and keep long running tasks alive without sitting in front of a single machine. + +The three companion releases stack with this. Programmatic access tokens make Codex callable from CI/CD without a human session. Hooks let you enforce policy on every Codex invocation, covering secret scanning, validators, logging, memory creation, and per directory behavior. HIPAA compliant local use opens healthcare workspaces for the first time. + +Together this is the package a platform team needs to actually ship Codex to a regulated company. It is not "AI for individual developers." It is AI inside the same compliance perimeter as the rest of your tooling. + +# What this means if you build with Codex + +The honest practical reads: + +* **If you are a solo developer**, mobile is the unlock. You stop being the bottleneck during the parts of your day you cannot be at a keyboard. Treat the phone as a steering wheel, not a workstation. +* **If you run a team**, Hooks and programmatic tokens are the items to evaluate this week. They are the difference between Codex being a personal tool and Codex being part of your pipeline. +* **If you work in a managed remote environment**, Remote SSH being GA means you can finally adopt Codex without arguing with your security team about local installs. +* **If you ship to regulated customers**, the HIPAA path is narrow but real. Worth a read if it applies to you. +* **If you were waiting for Windows**, OpenAI says support for connecting your phone to the Codex app on Windows is "coming soon." + +One thing the announcement does not change is that the model still needs a place to put the code it writes. Which brings us to the part most "AI shipped X" posts skip. + +# Ship more, worry about the backend less + +A faster, more available agent is only useful if there is something on the other side of its diffs ready to receive what it builds. Auth, databases, storage, functions, an API layer. The stack that usually eats the time the agent saves. If Codex finishes a feature in 20 minutes and you spend the next 4 hours wiring it into your backend, the gain is gone. + +[Appwrite](https://appwrite.io/) is an open source backend as a service that includes [Auth](https://appwrite.io/docs/products/auth), [Databases](https://appwrite.io/docs/products/databases), [Storage](https://appwrite.io/docs/products/storage), [Functions](https://appwrite.io/docs/products/functions), [Messaging](https://appwrite.io/docs/products/messaging), and [Sites](https://appwrite.io/docs/products/sites), an open source Vercel alternative for deploying your web app next to your backend. It is available as both managed Cloud and self hosted. Instead of standing up auth, schemas, file pipelines, deploys, and API layers yourself, you point Codex at an Appwrite project and the whole stack is one project. + +The loop is straightforward. Kick off a task from your phone on the way to the office. Codex inspects the repo, asks any clarifying questions, and starts building against an Appwrite backend that already exists. You review the diff at your desk. Pair the mobile anywhere workflow OpenAI just shipped with a backend that does not need to be hand built every time, and the session translates into product, not scaffolding. + +Want to try this loop end to end? [Create a free Appwrite project](https://cloud.appwrite.io/), open Codex on your phone, and ask it to wire up auth and a database against your new project. By the time you sit down at your desk, you have a working backend and a diff to review. + +Start with the [Appwrite quick start](https://appwrite.io/docs/quick-starts) for your stack, browse the [SDKs](https://appwrite.io/docs/sdks), or [join us on Discord](https://appwrite.io/discord) to share what you build. \ No newline at end of file diff --git a/static/images/blog/openai-just-shipped-codex-to-the-chatgpt-mobile-app/cover.avif b/static/images/blog/openai-just-shipped-codex-to-the-chatgpt-mobile-app/cover.avif new file mode 100644 index 0000000000000000000000000000000000000000..6ebc544a54480411823766a71de1b70a78df51c8 GIT binary patch literal 6391 zcmYj#Wmpv4^Y+rx-4Y6mbc&SHOG}8P)Y412EG^yL4FXGpgdp7=0@5Lk)FQEzbp6ri z`d#lkAI{9&_uS{4Ip59z0Dz~K&K^*(E5s7;X!a0GzQ3_O1pJtzKv=l{k3^8`?cv;DsW<#BxpvHNKDH~(r z6!gbb7y@zlca486`D0@7uk29`cID;!tN4H5BlAD-D+CIIK9Ug-sM*Iyj|X<;lko@s z(~?1)pca3X006oY0Dua442Yl*_rD&1_T13flLX=`r|b|wMf;E33ms!VjF(%>`lk8;X`I01|S z=;{Cxl$9mKru>)~_Wj2}Tm55*HpAV_}R6A*goH9ynl-K|#Sq%#$#U&>o ziPRIgE3pqanQa>8xD}jWj(<;P4T-=je0Z1^lA36FuJmv`ymlyThaWdodX>SyNg6hP zJ@$KD5FUb_`DAh}gTpPXa;A->^qF1Dgo_QP87mp3Xx4j+=s8dKg{f;a55tGxX(pU_ zJ6nj_)8>H9 z|0wxTE5Rb6Z{$g3B49}cMt_+VVpE*K|4#Xn=N#^upYbn)I}^Cjv0P6<{IfUqJd%11 zv|bQl`S)DJGPm80Xa}FAGFWjIO(5s@3(mbeZjtVGxod~NGVQm6o2iSa-hBLYw0BN4 zk~p{Pw)irAd1)}JaXn#jb7(M;SSv&OEBGW83Hn>Qha zNOZN4%G9f(tBeY8yRY&DBk)Cei=Ww2zjfHt$q0S_x(#H+BUdwO)GN!O5uuRKiL7oU z@(CT&KoRk2Ge$|TZ@{|Kl+-Jx%cPjkX9ibIpSHID1alSeP`%aDMxNzTxkc-yA3(Z$ zKuu>8j#z^MYK!j0{dY~B%MFSxU71phEP{N3`o@ln-}*l9Dy4tvt|S9So86U)HDu-N%eOoTkG!mFeW_wPn&^dFfETmU?*lV;79GD*#zy2s&Dbbk zr*DgF*3ab&Zb2nrnQ`Yr;L>qBgK%I>Nc+$dXiR=Q)SPs4vSyR2dl(KWPkcN#r$XQE2TN}gGYx1VC) z>!JR1`%JkKHC|yIKGFqhk)@t4qiB^LP*{a{+j~&A({mY~CeT-NLUJmL3t#3ImCdy* zk?(E~N-dW04Vix6scq1OC$3K8(gp!n&QUtJ^+5qK_J|M@#Wp^Q)sGQXf&p`*W5s&O zAN*KNw!+=o?TV@YiAl&crC0i8 zLPwd(t3MNLXxU2~BSlK{SJiI4-N1#Pz)CTp;*C+}p%am~a0+1^)~ZGEhMcr&mgBk? zH>DBhjVUy$NcO6H=gtv*=xz+#2JCoEgC83L9zK~dfOFge1|byZu!g-AVW@4W*X5>lu<&k!*R{v#wT;MlY!n z4&#sV3pX7K5dht~7tPXdNQ~%lTf-W1QUHR(Q$lOWpVw%4EEypfY}%<;b1lg$*VY_n~Hw zZ-}8#Fnv}2IQ7iWqS*02 zV^E$oqHv1rBo9m7#8h>bGtuTChC|Hlvo+sEjUU#icCuf)w5~;tEytTjyxzzAwp=k? zUQob#OwN&d?92AX5#zEE0Al#MEC-)@%9q)_ znKW-K?cnN_>vr>?5SZ|;`(tUw8P{WqA0;I5V{(gqe#s}I1*<#St-&gzGcuE z`UAN@sP^XriJ!QPyJR7RKA$A9ab7Dkl~MptSvd%wfbWC;)<*Fqh6LE#-`@;e@|vz~ z*6zDN>@O6Q-GXe4Z_0Zm9w?YBU^Fv{{WKzoV`K4RCaxfvXM|P&>6Ctuh0E3y`mqpU zZ8(DAd!bIi#F-d~RWwG!?G**LmXl7FegQfM;fvoA^|jX;IvUuPlg9mxd5JQ;7FUar zLj(0!>?}jh#<-Q8@hM^F# zN`O#paYC{bW$=X-zHI|f$H$VXU{GDfh1D(tO)DSLm6HI#(76qmFns;}!?SQO>P z_@)?&>`EcyT^1aPu}-hceD$8X44IOWn+EBiB@0rS?qonIq48MF-+Ru}E0~ZT(Qlc2 zK%;^y2$>@a*q&FsDEX9biNx~gv9K|m^~?XNBVmmnx(88?agCR=_U1?AXama2C^W4_ zxL2yIEX6}yfyF%3V?3vd?Xt}TTr1y06nYI4iA&?PO+r-t*>_@yOC04UoY@lV1q9qY!W_H(Fb)LM3Ec@ zr0@xSQ1K`q^WCup1Fb;Hr1~$mdi9qT=Il#O_Ot;O1PHrTgHMT(Z!v3_UI;IMv36K= zOYm;@7KiCpHncd4F^xoP!bAE;?bT)S>aqv6_Mkw7L>>!OI7{kSaNHUj$L`<{(iv{) zQX8AYPmbElWygxu5r!ELw>#tvhwp#f)D{1BW~SnjCi=W+VPJx9Z78LJZ!`Y9JD2|bb6{E+U^hq1-_8OyZGe(Od$X`85D?bTfL2l zve|62-4V`^+iP@^A|m4_?R~3up3du@|PZ&Tke=(x(FKZIIXIi(vf2+)v zg)SA{1|Gi|N(-;raSE{LU`{u{eOh))XI6}NgU)=1VZU1BxZ%7+mOpusgaE6Xrqm<_ z!MX0{Nt;1yuq1}=5Dw9jTc>*gtdBdKyAIAIBX(Hk_KX;l!n3npS_!3?1o8afo+x~h zHRokoWctSbY@Txcye~27nI-kB`OVafQ1%=^_bX37CaP@&1>1X}ErD^v+VRK*4q+h4 zp;2P+VZ1SjKTHid;`Y2fXk3N#rrM9ePjto0wi!b_-hPbcpf46^H*RU~H>P>Tn zj+y^jWiVAN`a-xJ#bt8djz+q>h9Ohu0hqkl5l*CLG}FfMjsNtrZ*=7~5a-!OaU=g} zxS7sO;)NsL?{~#it}5uYL;Z6>j0~m(v_QJ*h# zR5xTh0BRG6#1MXD9$O!S)7kUL7iCH&MNzLmQ{=#ErrOQ9S<0iXc1p9th%QO2vrSC; z_ry2geu9FpGivfk6`oMOEAaMl?FFwZgMinzJ}IQ>us|!eGnyS5`yf-a?D;;1!&86u zTm%lKJ4{t1*3pDJ5JvRw1XpDD6xvLF%3K;uY8 zMNMtWkDn>}-#%n$Luz)tV;4`<3tk(FG5!9J&4%ns#*cLcpN@Ru>_5NXhb;hIyL+7%r$iq{hHCo$)gkO5=vOD`9?-^qfM8@x|WnO*jgEAwh&Ek!wPEXtxnF(Ypc1c@g>&lN&U9a4rZtIBjB5`A<-KaM7V*NsPI>{-lc$H>bK1}gG9`UK0DVHs3cdk)D zpe?3z-t3oQlq(-nit#vd^d`gy*(DRJLg!$oRihDJ11%Nls*UpFcw9t90?gn&HbTnA zCe{ZC?MwcQzHe3&(iU^TV+TrYG=hXLoRyth?xNlUvSrxGsZuIhxAEpsQegQ_0|DAN z5hje~CALtOKEH}GE?=sd405d%PUV?rMR^-ih)lfVdf-7u>eNN&J01#6f3sW5`hMc; zsL{GnO_vNxTQ6EZ*$st7cg6%LbOoyLWvz9ND0I1=Zom)e8EAg+MOiF#Dv>@vP9a?M zDqnM`8~C_?C33FMsmm-Y|2>vAmPh_mjFyil!^l~Bpux;@p_Hg5b6nUYgB8>gsPjm+;UxjSaMRHtr+^HD&jR(^M}c~0?% z4>(e6EM#p2!)YPcI^&*lg|7vLhFytHqobH3ohqg#XTEA-*0^LmujIF~Ej zI3a{$k$Y4IzjLk-d0N;~SBcfT(J29vY^*G|^{S?X9e(L@DX-YPF${pm0BH{G1BG#IXi+r^BT&Kl@N8xr3M;6<`hTC=#=E4tjNh zk>lZJ76-p_$R|=o+wJilxZ2p4xAC!g9PN@BJydjGY}qdL31rVf26WQi+Aobf&(Hha zs>t%&{9MCng!=9$%^N#6*RMSe3-9FpLZAxd94>-?CdSQx)dx)cihhsdWi0A{<7GJi zIg0#yP5^)mKnnmIB+aHPfcicz=X>%FVoWp+eq$y$iLL$P<46CG+#C46GO1YiU$-1&9O~&}4Trm=jm>G?xkVbt! z;jW)Dxa>Q8P@e(Mc{XGW5i&2NCOQ4z)&AQt;eC!Pvo2@SJtHewfLd4Mtf`l~jdjz{ Xl?Dc_+F;QaK_9zcfCaY$GuHkO9=i6& literal 0 HcmV?d00001 From a8cbf10cbeb370a30a656db1694ab42e985c2cdf Mon Sep 17 00:00:00 2001 From: Aishwari Pahwa Date: Tue, 19 May 2026 17:38:45 +0530 Subject: [PATCH 2/8] Apply suggestion from @greptile-apps[bot] Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com> --- .../+page.markdoc | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/routes/blog/post/openai-just-shipped-codex-to-the-chatgpt-mobile-app/+page.markdoc b/src/routes/blog/post/openai-just-shipped-codex-to-the-chatgpt-mobile-app/+page.markdoc index 782cbd2220..4d4fe00ccf 100644 --- a/src/routes/blog/post/openai-just-shipped-codex-to-the-chatgpt-mobile-app/+page.markdoc +++ b/src/routes/blog/post/openai-just-shipped-codex-to-the-chatgpt-mobile-app/+page.markdoc @@ -94,7 +94,7 @@ One thing the announcement does not change is that the model still needs a place A faster, more available agent is only useful if there is something on the other side of its diffs ready to receive what it builds. Auth, databases, storage, functions, an API layer. The stack that usually eats the time the agent saves. If Codex finishes a feature in 20 minutes and you spend the next 4 hours wiring it into your backend, the gain is gone. -[Appwrite](https://appwrite.io/) is an open source backend as a service that includes [Auth](https://appwrite.io/docs/products/auth), [Databases](https://appwrite.io/docs/products/databases), [Storage](https://appwrite.io/docs/products/storage), [Functions](https://appwrite.io/docs/products/functions), [Messaging](https://appwrite.io/docs/products/messaging), and [Sites](https://appwrite.io/docs/products/sites), an open source Vercel alternative for deploying your web app next to your backend. It is available as both managed Cloud and self hosted. Instead of standing up auth, schemas, file pipelines, deploys, and API layers yourself, you point Codex at an Appwrite project and the whole stack is one project. +[Appwrite](/) is an open source backend as a service that includes [Auth](/docs/products/auth), [Databases](/docs/products/databases), [Storage](/docs/products/storage), [Functions](/docs/products/functions), [Messaging](/docs/products/messaging), and [Sites](/docs/products/sites), an open source Vercel alternative for deploying your web app next to your backend. It is available as both managed Cloud and self hosted. Instead of standing up auth, schemas, file pipelines, deploys, and API layers yourself, you point Codex at an Appwrite project and the whole stack is one project. The loop is straightforward. Kick off a task from your phone on the way to the office. Codex inspects the repo, asks any clarifying questions, and starts building against an Appwrite backend that already exists. You review the diff at your desk. Pair the mobile anywhere workflow OpenAI just shipped with a backend that does not need to be hand built every time, and the session translates into product, not scaffolding. From 3a6892abe9720175d4284bd46868f5a47826f221 Mon Sep 17 00:00:00 2001 From: Aishwari Pahwa Date: Tue, 19 May 2026 17:38:54 +0530 Subject: [PATCH 3/8] Apply suggestion from @greptile-apps[bot] Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com> --- .../+page.markdoc | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/routes/blog/post/openai-just-shipped-codex-to-the-chatgpt-mobile-app/+page.markdoc b/src/routes/blog/post/openai-just-shipped-codex-to-the-chatgpt-mobile-app/+page.markdoc index 4d4fe00ccf..4882025a57 100644 --- a/src/routes/blog/post/openai-just-shipped-codex-to-the-chatgpt-mobile-app/+page.markdoc +++ b/src/routes/blog/post/openai-just-shipped-codex-to-the-chatgpt-mobile-app/+page.markdoc @@ -100,4 +100,4 @@ The loop is straightforward. Kick off a task from your phone on the way to the o Want to try this loop end to end? [Create a free Appwrite project](https://cloud.appwrite.io/), open Codex on your phone, and ask it to wire up auth and a database against your new project. By the time you sit down at your desk, you have a working backend and a diff to review. -Start with the [Appwrite quick start](https://appwrite.io/docs/quick-starts) for your stack, browse the [SDKs](https://appwrite.io/docs/sdks), or [join us on Discord](https://appwrite.io/discord) to share what you build. \ No newline at end of file +Start with the [Appwrite quick start](/docs/quick-starts) for your stack, browse the [SDKs](/docs/sdks), or [join us on Discord](https://appwrite.io/discord) to share what you build. \ No newline at end of file From 2e93760af0c7bab40b2104d78587017d5fd446bf Mon Sep 17 00:00:00 2001 From: Aishwari Pahwa Date: Tue, 19 May 2026 17:39:22 +0530 Subject: [PATCH 4/8] Apply suggestion from @greptile-apps[bot] Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com> --- .optimize-cache.json | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/.optimize-cache.json b/.optimize-cache.json index bf19c09348..f7d1fc4999 100644 --- a/.optimize-cache.json +++ b/.optimize-cache.json @@ -872,7 +872,7 @@ "static/images/blog/open-source-backend-vs-managed-saas/cover.png": "5046fc4e3f3fa605c22d00ecb4b5156734c7d94000b49cae66a9e3bc49f81984", "static/images/blog/open-source-firebase-alternative/cover.png": "e0b98679795c00fd6d2d304b17273eaa6847bb1fc5706efa6cc3f3040ec6b4bb", "static/images/blog/open-source-regulated-environments/cover.png": "bbb57f0f8dc70f31923f34b5007f089a8539bdbca193c6f3a2594502bca551a2", - "static/images/blog/openai-just-shipped-codex-to-the-chatgpt-mobile-app/cover.png": "32abc0a488d7b7b8f2a4f5661801f415bb1cc1dc148768159bead6e5815c4bde", + "static/images/blog/openai-just-shipped-codex-to-the-chatgpt-mobile-app/cover.avif": "32abc0a488d7b7b8f2a4f5661801f415bb1cc1dc148768159bead6e5815c4bde", "static/images/blog/openjs.png": "4129893c1ea043b82af755a462fda8d78a7a5199638f84da549b2a2b29fbe976", "static/images/blog/oss-netlify-cover.png": "166487672a542d051b029ba285b703fc7f6c49d2fa27407f22c8f1848725120f", "static/images/blog/OSS-program.png": "b2cb73db872591b279e9b7ce87eb1c281732f394ef96a5d4e941ba3892924be7", From 537a73d6eea99c71ae068e1738a6fe1be306a656 Mon Sep 17 00:00:00 2001 From: Aishwari Pahwa Date: Thu, 21 May 2026 10:47:11 +0530 Subject: [PATCH 5/8] Update +page.markdoc --- .../+page.markdoc | 16 ++++++---------- 1 file changed, 6 insertions(+), 10 deletions(-) diff --git a/src/routes/blog/post/openai-just-shipped-codex-to-the-chatgpt-mobile-app/+page.markdoc b/src/routes/blog/post/openai-just-shipped-codex-to-the-chatgpt-mobile-app/+page.markdoc index 4882025a57..dc2e2592c9 100644 --- a/src/routes/blog/post/openai-just-shipped-codex-to-the-chatgpt-mobile-app/+page.markdoc +++ b/src/routes/blog/post/openai-just-shipped-codex-to-the-chatgpt-mobile-app/+page.markdoc @@ -28,15 +28,11 @@ If you have been running Codex as "a chat box that writes code," the second half Five things shipped, and the boundaries between them matter. -**Mobile access is now in preview on iOS and Android** through the ChatGPT app, across all plans including Free and Go. Before this release, Codex workflows were still primarily tied to desktop and IDE environments. - -**Remote SSH is generally available.** The desktop app auto detects hosts from your SSH config and lets you create projects and run threads inside remote machines just like you would locally. Before this, remote environment workflows existed, but they were far less integrated and not officially streamlined inside the Codex app. - -**Programmatic access tokens** provide scoped credentials issued directly from ChatGPT workspace settings, so Codex can be called from CI pipelines, release workflows, and internal automations without a human session. - -**Hooks are generally available** and can be used to scan prompts for secrets, run validators, log conversations, create memories, or customize Codex behavior for specific repositories and directories. - -**HIPAA-compliant local use** is supported for eligible ChatGPT Enterprise workspaces when Codex is used in local environments across the CLI, IDE, and App. +* **Mobile access is now in preview on iOS and Android** through the ChatGPT app, across all plans including Free and Go. Before this release, Codex workflows were still primarily tied to desktop and IDE environments. +* **Remote SSH is generally available.** The desktop app auto detects hosts from your SSH config and lets you create projects and run threads inside remote machines just like you would locally. Before this, remote environment workflows existed, but they were far less integrated and not officially streamlined inside the Codex app. +* **Programmatic access tokens** provide scoped credentials issued directly from ChatGPT workspace settings, so Codex can be called from CI pipelines, release workflows, and internal automations without a human session. +* **Hooks are generally available** and can be used to scan prompts for secrets, run validators, log conversations, create memories, or customize Codex behavior for specific repositories and directories. +* **HIPAA-compliant local use** is supported for eligible ChatGPT Enterprise workspaces when Codex is used in local environments across the CLI, IDE, and App. A few practical reads: @@ -100,4 +96,4 @@ The loop is straightforward. Kick off a task from your phone on the way to the o Want to try this loop end to end? [Create a free Appwrite project](https://cloud.appwrite.io/), open Codex on your phone, and ask it to wire up auth and a database against your new project. By the time you sit down at your desk, you have a working backend and a diff to review. -Start with the [Appwrite quick start](/docs/quick-starts) for your stack, browse the [SDKs](/docs/sdks), or [join us on Discord](https://appwrite.io/discord) to share what you build. \ No newline at end of file +Start with the [Appwrite quick start](/docs/quick-starts) for your stack, browse the [SDKs](/docs/sdks), or [join us on Discord](https://appwrite.io/discord) to share what you build. From 6a22520ce08a6d2dd540390686cb303ef8b2e63f Mon Sep 17 00:00:00 2001 From: Aishwari Pahwa Date: Thu, 21 May 2026 11:46:17 +0530 Subject: [PATCH 6/8] Add Appwrite plugin details for Codex Added information about the Appwrite plugin for Codex, including installation instructions and usage examples. --- .../+page.markdoc | 12 ++++++++++++ 1 file changed, 12 insertions(+) diff --git a/src/routes/blog/post/openai-just-shipped-codex-to-the-chatgpt-mobile-app/+page.markdoc b/src/routes/blog/post/openai-just-shipped-codex-to-the-chatgpt-mobile-app/+page.markdoc index dc2e2592c9..7db4800beb 100644 --- a/src/routes/blog/post/openai-just-shipped-codex-to-the-chatgpt-mobile-app/+page.markdoc +++ b/src/routes/blog/post/openai-just-shipped-codex-to-the-chatgpt-mobile-app/+page.markdoc @@ -97,3 +97,15 @@ The loop is straightforward. Kick off a task from your phone on the way to the o Want to try this loop end to end? [Create a free Appwrite project](https://cloud.appwrite.io/), open Codex on your phone, and ask it to wire up auth and a database against your new project. By the time you sit down at your desk, you have a working backend and a diff to review. Start with the [Appwrite quick start](/docs/quick-starts) for your stack, browse the [SDKs](/docs/sdks), or [join us on Discord](https://appwrite.io/discord) to share what you build. + +# The Appwrite plugin for Codex + +If you want to try this loop without wiring anything up yourself, we just shipped the [Appwrite plugin for Codex](https://appwrite.io/blog/post/announcing-appwrite-codex-plugin). One install gives Codex the Appwrite Docs MCP server and thirteen agent skills covering the CLI, every major SDK, and guided deploys for Sites and Functions. + +To install, run: + +```bash +codex plugin marketplace add appwrite/codex-plugin +``` + +Then start Codex, open `/plugins`, and confirm. Once active, try a prompt like "Scaffold a Next.js app that uses Appwrite for email/password auth and a todos table." Codex picks up the right skills, queries the docs MCP when it needs specifics, and generates code that uses real SDK calls instead of guessed ones. Kick that off from your phone on the way to the office, and there is a working diff waiting when you sit down. From d426cd1f2bd55856ea19ae71587e3358c63c534d Mon Sep 17 00:00:00 2001 From: Aishwari Pahwa Date: Thu, 21 May 2026 13:42:19 +0530 Subject: [PATCH 7/8] Update +page.markdoc --- .../+page.markdoc | 20 +++++++------------ 1 file changed, 7 insertions(+), 13 deletions(-) diff --git a/src/routes/blog/post/openai-just-shipped-codex-to-the-chatgpt-mobile-app/+page.markdoc b/src/routes/blog/post/openai-just-shipped-codex-to-the-chatgpt-mobile-app/+page.markdoc index 7db4800beb..03cf737d98 100644 --- a/src/routes/blog/post/openai-just-shipped-codex-to-the-chatgpt-mobile-app/+page.markdoc +++ b/src/routes/blog/post/openai-just-shipped-codex-to-the-chatgpt-mobile-app/+page.markdoc @@ -90,22 +90,16 @@ One thing the announcement does not change is that the model still needs a place A faster, more available agent is only useful if there is something on the other side of its diffs ready to receive what it builds. Auth, databases, storage, functions, an API layer. The stack that usually eats the time the agent saves. If Codex finishes a feature in 20 minutes and you spend the next 4 hours wiring it into your backend, the gain is gone. -[Appwrite](/) is an open source backend as a service that includes [Auth](/docs/products/auth), [Databases](/docs/products/databases), [Storage](/docs/products/storage), [Functions](/docs/products/functions), [Messaging](/docs/products/messaging), and [Sites](/docs/products/sites), an open source Vercel alternative for deploying your web app next to your backend. It is available as both managed Cloud and self hosted. Instead of standing up auth, schemas, file pipelines, deploys, and API layers yourself, you point Codex at an Appwrite project and the whole stack is one project. +[Appwrite](/) is an open source backend as a service that includes [Auth](/docs/products/auth), [Databases](/docs/products/databases), [Storage](/docs/products/storage), [Functions](/docs/products/functions), [Messaging](/docs/products/messaging), and [Sites](/docs/products/sites), an open source Vercel alternative for deploying your web app next to your backend. It is available as both managed Cloud and self-hosted. Instead of standing up auth, schemas, file pipelines, deploys, and API layers yourself, you point Codex at an Appwrite project and the whole stack is one project. -The loop is straightforward. Kick off a task from your phone on the way to the office. Codex inspects the repo, asks any clarifying questions, and starts building against an Appwrite backend that already exists. You review the diff at your desk. Pair the mobile anywhere workflow OpenAI just shipped with a backend that does not need to be hand built every time, and the session translates into product, not scaffolding. - -Want to try this loop end to end? [Create a free Appwrite project](https://cloud.appwrite.io/), open Codex on your phone, and ask it to wire up auth and a database against your new project. By the time you sit down at your desk, you have a working backend and a diff to review. - -Start with the [Appwrite quick start](/docs/quick-starts) for your stack, browse the [SDKs](/docs/sdks), or [join us on Discord](https://appwrite.io/discord) to share what you build. - -# The Appwrite plugin for Codex - -If you want to try this loop without wiring anything up yourself, we just shipped the [Appwrite plugin for Codex](https://appwrite.io/blog/post/announcing-appwrite-codex-plugin). One install gives Codex the Appwrite Docs MCP server and thirteen agent skills covering the CLI, every major SDK, and guided deploys for Sites and Functions. - -To install, run: +To make that pairing frictionless, we just shipped the [Appwrite plugin for Codex](https://appwrite.io/blog/post/announcing-appwrite-codex-plugin). One install gives Codex the Appwrite Docs MCP server and thirteen agent skills covering the CLI, every major SDK, and guided deploys for Sites and Functions. To install, run: ```bash codex plugin marketplace add appwrite/codex-plugin ``` -Then start Codex, open `/plugins`, and confirm. Once active, try a prompt like "Scaffold a Next.js app that uses Appwrite for email/password auth and a todos table." Codex picks up the right skills, queries the docs MCP when it needs specifics, and generates code that uses real SDK calls instead of guessed ones. Kick that off from your phone on the way to the office, and there is a working diff waiting when you sit down. +Then start Codex, open `/plugins`, and confirm. From there, the loop is straightforward. Kick off a task from your phone on the way to the office with a prompt like "Scaffold a Next.js app that uses Appwrite for email/password auth and a todos table." Codex picks up the right skills, queries the docs MCP when it needs specifics, and generates code that uses real SDK calls instead of guessed ones against an Appwrite backend that already exists. You review the diff at your desk. Pair the mobile anywhere workflow OpenAI just shipped with a backend that does not need to be hand built every time, and the session translates into product, not scaffolding. + +Want to try this loop end to end? [Create a free Appwrite project](https://cloud.appwrite.io/), install the plugin, open Codex on your phone, and ask it to wire up auth and a database against your new project. By the time you sit down at your desk, you have a working backend and a diff to review. + +Start with the [Appwrite quick start](/docs/quick-starts) for your stack, browse the [SDKs](/docs/sdks), or [join us on Discord](https://appwrite.io/discord) to share what you build. From 3f9224907e2348a3017973615c5099ef6b9b27c0 Mon Sep 17 00:00:00 2001 From: Aditya Oberai Date: Thu, 21 May 2026 21:19:56 +0530 Subject: [PATCH 8/8] Apply suggestion from @adityaoberai --- .../+page.markdoc | 2 -- 1 file changed, 2 deletions(-) diff --git a/src/routes/blog/post/openai-just-shipped-codex-to-the-chatgpt-mobile-app/+page.markdoc b/src/routes/blog/post/openai-just-shipped-codex-to-the-chatgpt-mobile-app/+page.markdoc index 03cf737d98..0a6456b9d1 100644 --- a/src/routes/blog/post/openai-just-shipped-codex-to-the-chatgpt-mobile-app/+page.markdoc +++ b/src/routes/blog/post/openai-just-shipped-codex-to-the-chatgpt-mobile-app/+page.markdoc @@ -101,5 +101,3 @@ codex plugin marketplace add appwrite/codex-plugin Then start Codex, open `/plugins`, and confirm. From there, the loop is straightforward. Kick off a task from your phone on the way to the office with a prompt like "Scaffold a Next.js app that uses Appwrite for email/password auth and a todos table." Codex picks up the right skills, queries the docs MCP when it needs specifics, and generates code that uses real SDK calls instead of guessed ones against an Appwrite backend that already exists. You review the diff at your desk. Pair the mobile anywhere workflow OpenAI just shipped with a backend that does not need to be hand built every time, and the session translates into product, not scaffolding. Want to try this loop end to end? [Create a free Appwrite project](https://cloud.appwrite.io/), install the plugin, open Codex on your phone, and ask it to wire up auth and a database against your new project. By the time you sit down at your desk, you have a working backend and a diff to review. - -Start with the [Appwrite quick start](/docs/quick-starts) for your stack, browse the [SDKs](/docs/sdks), or [join us on Discord](https://appwrite.io/discord) to share what you build.