Apple is integrating advanced agentic coding capabilities into its Xcode development environment. On Tuesday, the company announced the launch of Xcode 26.3, which will enable developers to utilize agentic tools, including Anthropic’s Claude Agent and OpenAI’s Codex, directly within Apple’s official app development suite.
The Release Candidate for Xcode 26.3 is currently available to all Apple Developers via the developer website, with a broader release to the App Store slated for a later date.
This significant update follows last year’s Xcode 26 release, which initially introduced support for ChatGPT and Claude within Apple’s integrated development environment (IDE). This IDE is crucial for developers creating applications for iPhone, iPad, Mac, Apple Watch, and other Apple hardware platforms.
The incorporation of agentic coding tools allows AI models to leverage a greater range of Xcode’s functionalities, enabling them to complete tasks more effectively and automate complex development processes.
These models will also have access to Apple’s most current developer documentation, ensuring their code utilizes the latest APIs and adheres to recommended best practices during development.
Initially, these agents can assist developers by exploring project structures, understanding metadata, building projects, running tests to identify errors, and subsequently fixing any detected issues.
To facilitate this launch, Apple collaborated closely with both Anthropic and OpenAI to refine the user experience. The company specifically focused on optimizing token usage and tool calling to ensure the agents operate efficiently within Xcode.
Xcode now employs MCP (Model Context Protocol) to expose its capabilities to these agents and connect them with its internal tools. This means Xcode can now integrate with any external MCP-compatible agent for various functions, including project discovery, code modifications, file management, previews, code snippets, and access to the most up-to-date documentation.
Developers interested in experimenting with the agentic coding feature should first download their desired agents from Xcode’s settings. They can also link their accounts with the AI providers by logging in or entering their API key. A convenient drop-down menu within the application enables developers to select their preferred model version (e.g., GPT-5.2-codex versus GPT-5.1-mini).
Through a prompt box on the left side of the screen, developers can instruct the agent on the type of project to build or the code changes to implement using natural language commands. For instance, they could direct Xcode to add a specific feature to their app, detailing how it should appear and function, utilizing one of Apple’s provided frameworks.
As the agent begins its work, it breaks down tasks into manageable steps, providing clear visibility into the process and code modifications. It also proactively searches for necessary documentation before starting to code. Code changes are visually highlighted, and a project transcript on the side of the screen offers developers insights into the agent’s underlying actions.
Apple believes this level of transparency will particularly benefit new developers learning to code. To support this, the company is hosting a “code-along” workshop on Thursday on its developer site, allowing users to observe and learn how to use agentic coding tools in real-time alongside their own Xcode instance.
Upon completion of its process, the AI agent verifies that the generated code functions as intended. Armed with test results, the agent can then iterate further on the project if necessary to correct errors or address other issues. (Apple advises that prompting the agent to thoroughly plan its approach before writing code can sometimes improve the overall development process.)
Furthermore, if developers are dissatisfied with the outcomes, they can easily revert their code to a previous state at any point, as Xcode automatically creates milestones with every change made by the agent.