Hello, developers! Xcode 26.3 unlocks the power of agentic coding! Get ready to explore new features that released last Tuesday. If you haven't already, you can download Xcode 26.3 from the Apple Developer Downloads page Check out the official documentation for getting started and how to configure agents in Xcode: ‎ Writing code with intelligence in Xcode Generate code, fix bugs fast, and learn as you go with intelligence built directly into Xcode https://developer.apple.com/documentation/xcode/writing-code-with-intelligence-in-xcode/ ‎ Setting up coding intelligence Enable third-party coding tools that you want to use in the coding assistant https://developer.apple.com/documentation/xcode/setting-up-coding-intelligence ‎ Giving external agentic coding tools access to Xcode Let agentic coding tools access your project and Xcode capabilities using the Model Context Protocol https://developer.apple.com/documentation/xcode/giving-agentic-coding-tools-access-to-xcode ‎ Code-along: Experiment with coding intelligence in Xcode 26 Presented live at a Meet with Apple event: Learn how coding intelligence features can help you write code and fix errors with example prompts and resources. ‎ ‎We are also introducing a new tag to the forums Coding intelligence ✨. Use this tag when discussing these new features to increase visibility on the forums to help others chime in.

Codex will no longer be available in Xcode 26.3 starting on Friday, May 8th. Please upgrade to Xcode 26.4 or later to continue using Codex.

It doesn't show any files, only warning or errors for me. Same behavior with @

I am using Coding Intelligence, powered by Claude, in the first production release of Xcode 27.0 (in MacOS 26.7). Initially it was working ok. But after less than a day, it now crashes every time it needs to ask for permission for any shell command. And each time I restart things, it burns through a ton of tokens. ☹️ Any suggestions? (I'll file a bug report right now…)

Bug Report: Xcode 26.6 Codex Agent Fails with Response Decoding and Model Compatibility Errors Severity: High — Core agent functionality unavailableComponent: Xcode Intelligence / Codex Agent IntegrationEnvironment: macOS, Xcode 26.6 (17F113)Date reported: November 1, 2026 Summary The built-in Codex coding agent in Xcode consistently fails to process requests, despite successful ChatGPT Pro authentication and a functioning standalone Codex CLI. The problem occurs even with trivial prompts and prevents use of Codex's autonomous development functionality, including the Pursue Goal workflow. The standard ChatGPT integration inside Xcode works correctly. The failure is isolated to the Codex agent integration. Environment ComponentDetailsIDEXcode 26.6 (17F113)Built-in Codex agent0.140.0 (version observed before reinstall)Standalone Codex CLI0.166.0ACP adapter testedcodex-acp 2.5.0AccountChatGPT ProAuthenticationChatGPT account authenticationAffected projectsMultiple SwiftUI projectsModel testedgpt-5.5, gpt-6.1-solActual Behavior When initiating a conversation with the built-in Codex agent in Xcode, the request fails with the following messages: Error 1: Response decoding failure The data couldn’t be read because it isn’t in the correct format.Accompanied by: Your request couldn't be completed.Error 2: Unsupported model With the original model selection: Codex encountered an error: The 'gpt-5.5' model is not supported when using Codex with a ChatGPT account.After explicitly configuring gpt-6.1-sol: Codex encountered an error: The 'gpt-6.1-sol' model is not supported when using Codex with a ChatGPT account.Error 3: ACP communication failure When testing a separately registered Codex ACP adapter: JSON-RPC request with ID [...] failed to produce a response.This was observed in a separate integration experiment and should not automatically be assumed to share the same root cause as the built-in Codex failure. Expected Behavior The built-in Codex agent should: Authenticate with an eligible ChatGPT Pro account. Select a model supported by the account and installed agent version. Correctly initialize the agent and decode its protocol responses. Execute basic prompts without errors. Support Xcode's native agentic development workflows, including Pursue Goal. Present actionable diagnostics when authentication, model selection, or protocol initialization fails. Reproduction Steps Open Xcode 26.6. Open an existing SwiftUI project. Open the Coding Assistant. Start a new conversation. Select Codex under Agents, rather than OpenAI under Chat. Submit a trivial prompt, such as: Reply with exactly OK. Observe the request failing with a response-format error or an unsupported-model error, depending on configuration. The issue is reproducible across multiple attempts and is not specific to a complex coding task. Troubleshooting Performed The following troubleshooting actions have already been attempted: Reauthenticated using the ChatGPT account. Tested configuration-file-based authentication. Reset Codex authentication. Reinstalled Xcode. Reinstalled the Xcode Codex component. Restarted Xcode and created new conversations. Tested explicit model selection through config.toml. Tested standalone Codex with the same ChatGPT Pro account. Tested a separate ACP adapter integration. The built-in Codex agent remains unusable. Confirmed Successful Control Test Standalone Codex CLI 0.166.0 successfully executed a request using GPT-6.1 Sol. Command: codex exec --model gpt-6.1-sol --skip-git-repo-check \ "Reply with exactly OK. Do not edit files or use tools."Observed output: OpenAI Codex v0.166.0 model: gpt-6.1-sol provider: openai reasoning effort: high codex OKThe request completed without an API error. Additionally, Xcode's separate ChatGPT integration displays the ChatGPT Pro account as authenticated, with usage status Under Limit. These observations strongly suggest that the problem is specific to the Xcode Codex integration or its runtime environment, rather than a general account authentication failure. Suspected Root Causes The following are hypotheses requiring investigation, not confirmed causes: 1. Bundled agent version incompatibility Xcode was observed using Codex 0.140.0, while standalone Codex 0.166.0 successfully accessed GPT-6.1 Sol. Please verify whether the bundled Codex agent is compatible with currently available ChatGPT-authenticated models. 2. Model discovery or selection mismatch The bundled agent's model catalog included models that were subsequently rejected by the service. Please investigate whether Xcode refreshes the model catalog, respects model configuration overrides, and selects models based on account eligibility. 3. Agent protocol response decoding The generic data-format error suggests a possible failure to decode an agent response, handle an unsupported response schema, or propagate an underlying API error. Please inspect the agent initialization, response parsing, and error-handling paths. 4. Configuration and authentication isolation The normal ChatGPT integration authenticates successfully, while the Codex agent fails. Please verify how Xcode manages distinct authentication contexts, configuration locations, agent versions, and model capabilities. Requested Engineering Actions Please investigate and address the following: Ensure the Xcode Codex agent is compatible with supported ChatGPT-authenticated models. Verify that the bundled agent version is current and that component updates are correctly delivered. Validate model availability before dispatching requests. Ensure the model selected in Xcode matches the actual model submitted by the agent. Correct any incompatible protocol or JSON response handling. Expose the underlying error details rather than only reporting a generic data-format failure. Confirm that reinstalling or updating the Xcode Codex component restores a functional agent without requiring manual configuration changes. Add regression coverage for ChatGPT Pro authentication, model selection, agent initialization, and basic prompt execution. Impact The issue prevents me from using the Codex agent directly inside Xcode, despite having a ChatGPT Pro subscription and a fully functional standalone Codex CLI. As a result, I cannot use the integrated agentic coding experience or the Pursue Goal workflow that previously worked. Please investigate the bundled Codex runtime, model selection, and protocol compatibility. The successful standalone Codex 0.166.0 test provides a control case for reproducing and isolating the failure.

This was maddening: I had written a very long and detailed prompt, then went to select "Codex" from the drop down menu at the top "start a new conversation", and it wiped out the entire prompt I was working on. Well, of course it did: that drop down is part of the "Start a new Conversation" icon button, but it's position is too far away from the actual icon to infer that is its purpose. It should be a popup that shows the current choice for model (GPT-5, GPT 4.1, Codex). ANOTHER lost prompt problem. If you have a prompt-in-progress (AI is maybe building some code), and Xcode crashes, it might not save that prompt for when you re-launch. And how might that happen? Well, there's still numerous ways using the coding assistanty might crash. One I've notice is if you switch out to another app, or if you invoke a screen capture, it might crash.

Coding "Intelligence" < Stable IDE I don't use Code Intelligence. I don't use AI. I can disable it in settings, but the button and the hints are always there. Xcode 27 vs. Xcode 16 feels like a dumbed down code slot machine for minors, not a serious product for serious programmers with it's big shiny buttons and buttons to "Generate", "Start a new Chat", etc. If you want to make toys for people to pretend like they are making software — fine. But leave the product for actual programmers alone. I don't want "AI Slop" buttons. I want to write code BY HAND and the IDE to be stable and fast.

I can't use cloud-based AI providers, so I'm very excited about the potential of local models with Xcode's Coding Assistant. I've been playing around with a local MLX model, and have been generally pretty satisfied with it. But I'm curious - what features am I missing out on by using a local model instead of a larger, cloud-based model? Is Xcode's ability to run tests, build the project, etc. limited by the use of a local model? Does it limit the ability to do more complex agentic workflows? If so, how? Basically, local models in Xcode seem to be working pretty well for me, and I'm curious to know what cloud-based models can offer that I don't already have.

I'm trying to use Xcode Coding Assistant connected to local AI server, but the chat fails after 5 minutes with error: The request timed out. Detail: No Data/Bytes for request: http://localhost... AI server gives me info that client disconnected: 2026-10-19 22:24:23 [INFO] [nvidia/nemotron-3.5-super-vl] Prompt processing progress: 88.3% 2026-10-19 22:24:30 [INFO] [LM STUDIO SERVER] Client disconnected. Stopping generation... (If the model is busy processing the prompt, it will finish first.) 2026-10-19 22:24:30 [INFO] [nvidia/nemotron-3.5-super-vl] Finished streaming response 2026-10-19 22:24:31 [DEBUG] 17.12.133.523 W srv stop: cancel task, id_task = 0 2026-10-19 22:24:37 [DEBUG] 17.18.441.198 I slot release: id 0 | task 0 | stop processing: n_tokens = 47985, truncated = 0 It looks like http/tcp default timeout. AI suggests that it should be possible to set Xcode Coding Assistant timeout, but I cannot find it anywhere. How is Xcode Coding Assistant supposed to work if it shuts down on default network protocol timeout? I haven't tried yet on nVidia hardware, which should be faster, but it could still be a problem when working on multiple files in project with inference running on one Blackwell chip. I do not have nVidia rack with 72 chips, it wouldn't fit to my home office. How do I change the network timeout for Coding Assistant? Configuration: MacOS 27.0.1 (26A434) Xcode 27.0

Installed from the Mac App Store Signed in with my Apple ID in Xcode Apple Intelligence is enabled Chat providers work normally (ChatGPT, Claude, Google) What I expected According to Apple's documentation, the Intelligence settings should display built-in Agents such as: Claude Agent Codex with a Get button to install them. Reference: https://developer.apple.com/documentation/xcode/setting-up-coding-intelligence What actually happens After upgrading to Xcode 26.6, the Agents section only contains: Allow external agents to use Xcode tools Add an Agent... The built-in Agents (Claude Agent and Codex) are completely missing.

If you're on an Intel Mac, here's the reason and the fix. Cost me an evening, and I found nothing online except closed bug reports with the same symptom. The problem: The browser flow works fine. You authorize, the page confirms, Xcode acts like nothing happened, no error anywhere. Xcode only downloads the Apple Silicon (arm64) build of its agent binaries. On Intel they can't run at all. The process meant to receive your login dies instantly with "bad CPU type in executable" and Xcode never surfaces it. Check if this is you: file ~/Library/Developer/Xcode/CodingAssistant/Agents/claude/*/claude If it says arm64 and your Mac is Intel, that's the problem. The fix: Open the Info.plist next to the binary. It has the exact download URL. Download the x64 build of the same version. → Claude: change darwin-arm64 to darwin-x64 in that URL. → Codex: grab codex-x86_64-apple-darwin.tar.gz from the matching rust-v release on GitHub. Replace the binary and make it executable: chmod 755 Update the checksum in Info.plist. It's a plain SHA-512: shasum -a 512 Restart Xcode and sign in again. Bonus (Claude Pro / Max) Run claude setup-token in Terminal, then add the token to: ~/Library/Developer/Xcode/CodingAssistant/ClaudeAgentConfig/settings.json {"env": {"CLAUDE_CODE_OAUTH_TOKEN": "your-token"}} Pick "Authenticate with Configuration File" in the agent menu. Runs on your subscription no API key. Heads up: an Xcode update can re-download the wrong binary, so you may have to repeat this until Apple fixes the architecture detection.

Dear Xcode Team, Please consider adding an option to position the Coding Assistant on the right side of the editor. Currently, when the assistant panel is open, it makes it difficult to access the Project Navigator at the same time. Having the ability to move the Coding Assistant to the right side would allow developers to keep the Project Navigator visible while interacting with the assistant, improving workflow and efficiency. Thank you

In Xcode's Settings → Intelligence section, there's no built-in AI agent/model available, only chat bots. I'm not able to download a model through the Components section as well, but no models are offered or downloadable there either. Environment: Xcode 26.6 and Xcode 27 beta 6 macOS 26.6.2 MacBook Pro (Apple M4 Pro)

When will this be available? (Opencode + LMStudio + Qwen 3.6 27B) FB23978071

Nothing changed on my end, Pro subscription still active, but Xcode 26.5 (17F42) simply does not want to log into Codex anymore. It opens the OAuth flow and after finishing and closing the window, it still says "Not Signed In". Codex login works everywhere else, including the Codex app. What to do?

In the video "Xcode, Agents and You" (direct link), the presenter asks for a walkthrough of the open project and Xcode presents a rich-text overview of the app's architecture, with color-coded diagrams and clickable links to the files it discusses. I can't get that to work when I try it, though. Is this a preview of a feature that isn't yet available in the first beta of Xcode 27? Is this because I'm using a local model or because it only works with a specific kind of model?

I'm unable to use the Claude Agent in Xcode 26 (stable release) running on macOS 27 beta 5. Steps to reproduce: Open Xcode 26 Settings > Intelligence Select Claude Agent Expected: The Account section loads and allows me to sign in / manage my Claude account. Actual: The Account row shows a loading spinner indefinitely and never resolves, so I can't sign in or use the Claude Agent at all. Environment: Xcode 26 (stable), macOS 27 beta 5. Is this a known incompatibility with macOS 27 beta 5? Is there a workaround (e.g. clearing a cache or re-adding the account) to get the Claude Agent account section to load?

I've been trying for months to get Gemini 3.5 Flash to work with Xcode 27 Beta (1-3) agentic coding but have failed. Xcode currently uses Gemini CLI v0.42.0 from last year, so it can't call Gemini 3.5 Flash even though all that needs to change is a few characters in the API call. Gemini users are currently forced to choose between 3.1 Pro and 3.1 Flash Preview--both weren't the best choices two months ago, but with the release of 3.6 Flash which is better at coding that 3.1 Pro and miles cheaper, these choices are now unsuitable for any serious workload. I live in Hong Kong, where OpenAI and Anthropic services aren't offered, so Google is the only (native) AI provider for Xcode agentic coding. The cost issue from funneling all the work into 3.1 Pro is terrible especially with some agent bugs that have yet to be ironed out. Given Apple's close partnership with Google on AI, I'm genuinely surprised that after 4 Betas we're still using a Gemini CLI from 2025 that forces users to choose between two very unappealing models. Xcode is basically forcing Google users to switch to OpenAI/Anthropic. Questions: I'd like to know if this will be fixed soon and whether Google users should expect slower updates (i.e. be forced to use older models). Does anyone know of any workarounds?

I’m seeing a reproducible issue in both Xcode 26 and Xcode 27 where Coding Assistant appears to stop autocompleting when a project contains a larger number of Swift Package dependencies. Symptoms: Autocomplete suggestions from Coding Assistant stop appearing. File targeting with @ stops working. The issue typically appears shortly after opening the project If I start a brand-new chat immediately after opening the project, Coding Assistant sometimes works briefly before failing Reducing or removing Swift Package dependencies fixes the issue I’ve created a sample project that reproduces the problem and have also filed Feedback Assistant report: FB23341243 The issue seems related to package count or project complexity rather than any specific package, although I haven’t yet identified the exact threshold that triggers it. Has anyone else run into this or has anyone found a workaround?

Accepting a proposal in chat switches windows and no code change applied. (hoping part of issue might be chat window takes over right side rather than swapping with navigator like before in previous Xcode 26.5

Never have this problem running in the first version of Xcode 27 until download the latest Xcode 27. At least "Fable" should be added in the dropdown to give a sense that it could be selected and be more careful (in dropdown only have "Sonnet", "Opus", "Sonnet+Opus" and more) I also submit Feedback request FB23973503 with screen capture & more Xcode Claude AI Investigation details

As of May 8th, 2026, after I upgraded to "codex-cli 0.129.0" via Xcode, I can no longer sign into my ChatGPT account in Xcode Intelligence Settings. The Log In window is always successful, but the settings never update to show the login being successful. It's a constant loop. Perhaps a hand-shake fail. Versions: Tahoe 26.3.1, Xcode 26.4.1, codex-cli 0.129.0.