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.

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.

How do I select between GPT and Claude, and the model, when opening a new task for the coding agent with the ‘New Conversation’ button in the toolbar?

I set up Gemini in Xcode 27 beta in Settings using an API key from Google AI studio. When I select "New Conversation" and "Gemini" and type "hi" it just spins forever (many minutes) - no reply or error message. Is there any way to figure out what's wrong?

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 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?

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?

Is anyone else finding model switching a bit disruptive during longer coding sessions? As we start using different models for exploration, implementation and review, having to go through settings interrupts the workflow. I would love to see model and reasoning controls directly inside the Coding Assistant UI.

Environment: macOS: Tahoe 26.4 Xcode: 26.5 ChatGPT account successfully authenticated Codex CLI installed and working Codex App installed and working Steps to reproduce: Install Xcode Codex/OpenAI component version where GPT-5.4 is the highest available model. Sign in with ChatGPT account. Confirm Codex works in Xcode. Update the OpenAI/Codex component in Xcode. Attempt to sign in again. Expected result: Xcode completes authentication and shows signed-in status. Actual result: Browser successfully completes authentication and displays "Signed in to Codex." Xcode remains in a sign-in loop or shows "Not Signed In." Authentication never completes. Additional observations: Codex CLI authentication succeeds. Codex App authentication succeeds. ChatGPT web/app access succeeds. Anthropic/Claude component updates continue to work normally in Xcode. Issue appears specific to the updated OpenAI/Codex Xcode component. Older component (GPT-5.4 maximum model available) authenticates successfully.

It is not possible to log into my Claude account in Xcode 27 beta 1 for the coding intelligence chat. When I click "Sign In," the browser opens, I authorize, the browser closes, and that's it. Nothing happens. Xcode does not recognize the login. I'm on MacOS 26.4.1 (25E253) and using Xcode 27.0 beta (27A5194q)

Has anyone managed to get Copilot working using the new ACP integration in Xcode 26.6 and Xcode 27 beta? I have tried setting it up using the path to the copilot binary as instructed and the flags --acp --stdio but Xcode just displays a "Your request couldn't be completed" error: JSON-RPC global stream failed: The operation couldn’t be completed. (IDEIntelligenceProtocol.JSONRPCElement.Error error 1.) I am running the latest version of copilot cli and it works in many other IDEs.

Since today (5th June), my Codex seems to be broken. First there is no response (wait for hours), now it just crash with this error whenever I issue a command: JSON-RPC global stream failed: The operation couldn’t be completed. (IDEIntelligenceProtocol.JSONRPCElement.Error error 1.) I have a ChatGPT Pro subscription and I don't know what to do. I have restarted Xcode, sign out and in of my account, delete DerivedData, restart machine, reinstall Xcode.

I gave Xcode 27 Beta 1 + Claude Agent my app codebase and set it to work on reviewing the SwiftUI related codebase and suggest improvements in the areas view performance and SwiftUI conformance. I am extremely happy with the results I got. From my experience the Xcode backed Claude Agent is far more versed in SwiftUI related task than pure Claude Code or Claude Code + SwiftUI skill extensions. I was able to visibly increase the performance of several complex views in my app. In addition the Xcode backed agent solved a Liquid Glass related bug with view tinting that I was not able to solve with Claude Code before on its first try. However, in Beta 1 Claude Agent BURNS through tokens at a mind boggling speed. I perceive the agent as a bit to verbose in telling me what it is doing. Allowing the agent to start a simulator and use the app for exploration is setting tokens on fire. I hope that we see refinements in these areas before Xcode 27 is released.

I'm trying to use Codex via the Xcode interface, and I can not find a way to set the reasoning level. There are no settings to be found anywhere. What am I missing? I mean, I can continue to use Codex's own app, but I was hoping to use Xcode's new agentic interface, which seems superior to v26 and somewhat more useful. But if it's going to default to medium and I have no control over it, of course, I would continue using Codex's app rather than Xcode. What's weird, I get responses like this when I ask about: What's your current reasoning level and how can I adjust it? This session appears to be using the default reasoning level: medium. You generally adjust it outside my reply, in the model/session settings for Codex or the API request. Typical options are: • low: faster, cheaper, best for simple edits or direct questions • medium: default balance • high: better for complex debugging, architecture, tricky code changes • xhigh: deepest reasoning, slower, useful for hard multi-step work Inside this chat, I can’t change my own base reasoning level mid-response. You can still tell me “use high reasoning for this” and I’ll treat the task more deliberately, but the actual runtime setting has to be changed in the client/session configuration. So, it doesn't even know the level it's using or how to adjust.

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 attempting to set up Gemini agentic support in Xcode 26.6 RC (and I'm assuming the process is the same in 27), but since I need to use Gemini with a corporate account, I have a Google Cloud Project ID, not an API key, so it seems like the configuration file route is the right path? After consulting Gemini and filling out the .env file and settings.json files, Xcode is complaining about needing to be authenticated, and I can’t seem to see how to trigger the auth, like one might do in Gemini CLI. I'd appreciate any suggestions or advice at this point. Thanks in advance!

Hello, I’m trying to use a third-party MCP server (ICM) in Xcode 27’s new Coding Assistant. The server appears to start correctly and shows up when I run /mcp, but as soon as an agent tries to use it, Xcode appears to rewrite my config and changes "enabled = true" to "enabled = false". My configuration looks like this: [mcp_servers.icm] command = "/opt/homebrew/bin/icm" args = [ "serve", "--no-embeddings", "--db", "/Users/Sam/emdash/worktrees/Castle/emdash/custom-widget-system-5dubi/.icm-xcode/memories.db" ] enabled = true I’ve spent quite a while debugging this and found some interesting behaviour: The MCP server appears when I type /mcp. SQLite databases can be created and written inside the current workspace. SQLite databases can also be created and written in /tmp. SQLite databases outside the workspace (for example in ~/Library/Application Support) fail with readonly/open errors. Ordinary file operations outside the workspace still work fine. What I’m trying to understand is: Why is Xcode disabling the MCP server? What filesystem access should MCP servers have? Are they intentionally restricted to the workspace and /tmp? Is there a supported way to allow an MCP server to access data stored elsewhere on disk? Has anyone else run into this? Thanks!

Could you please add support for https://agents.md in Xcode coding intelligence? By default or at least optionally add the AGENTS.md file to coding agent sessions.

How to select GPT-5.5 in Chat? In Xcode 27 beta under Codex Settings I see GPT 5.5. But in "New Conversation" -> Chat / OpenAI there's only the old GPT 5 and 4.1!?

Hello there everyone. I've got numerous support projects using old school development. Things from Objective-C to RxSwift and UIKit. I've tried the LLM/Agents and they just destroy the codebase over time, so I have given up. The final straw was trying to get it to fix a POSTCODE validation issue which was using some novel ways to provide validation in place using RxSwift. Has anyone had any luck with these LLMs on Legacy codebases? Would love to hear tips.

We would like to use Xcode's new agentic coding feature (probably with Claude Agent). I had a look at the privacy statement, and under "Third-Party Agentic Coding Tools" it says the following: Capabilities include but are not limited to code searching, retrieving build logs, building source code, and arbitrary code execution. This is all expected, except for the bit about "arbitrary code execution". What does this mean, exactly? This sounds like a backdoor to my machine, surely that's not it? Does "code" refer to the code in my project, or any code the agent may provide? I know that other agentic coding tools may execute terminal commands, which you have to (or can select to) confirm first. Is that what is meant by arbitrary code execution? How arbitrary are we talking? And can I configure Xcode to always ask my permission first before any code is executed?