Apple is adding support for agent coding in Xcode 26.3, allowing developers to create apps using tools like Anthropic’s Claude Agent and OpenAI’s Codex directly in Xcode.
Agenttic coding allows Xcode to autonomously perform more complex app development tasks. Claude, ChatGPT, and other AI models have been available in Xcode since Apple added intelligence capabilities in Xcode 26, but until now the AI had limitations and was unable to take actions on its own. This depends on your option to use an AI coding assistant.
AI models have access to more Xcode features to work toward project goals, and Apple worked directly with Anthropic and OpenAI to configure agents for use with Xcode. Agents can create new files, explore the project structure in Xcode, build the project directly and run tests, take image snapshots to double-check their work, and access complete Apple developer documentation designed for AI agents.
Adding agents to Xcode is a one-click experience in Xcode settings, and agents are automatically updated when AI companies release updates. To use these coding tools with Xcode, developers must set up an Anthropic or OpenAI account and pay based on API usage.
Apple says it aims to reduce token usage and make Claude Agent and Codex run more efficiently. Easily interchangeable between agents within the same project, giving developers the flexibility to choose the best agent for a particular task.
Apple used OpenAI and Anthropic for Xcode integration, but the features in Xcode 26.3 can be used with any agent or tool that uses the open standard model context protocol. Apple has released documentation to help developers configure and connect the MCP agent to Xcode.
Developers can use natural language commands to instruct AI agents to complete projects, such as adding new features to an app. Xcode then works with the agent to break down the instructions into smaller tasks, from which the agent can work independently. Here’s how the process works:
- Developers ask integration agents to add new functionality to their apps.
- The agent examines the current project to see how it is configured.
- The agent checks all relevant documentation for code snippets, code samples, and updated APIs.
- The agent starts working on the project and adds code as it goes.
- The agent builds the project and uses Xcode to verify its operation.
- If there are any errors or warnings, the agent continues to run until all issues are resolved. You can access the build log and modify your project until it’s perfect.
- The agent finishes with an overview of everything that happened to give the developer a clear picture of the implementation.
In the project sidebar, developers can track agent behavior using transcripts or click to see where code was added. Developers can always go back to before an agent or model made a change, giving them the option to undo undesired results or try multiple options to introduce new functionality.
According to Apple, agent coding allows developers to simplify their workflows, make changes faster, and bring new ideas to life. Apple also sees this as a learning tool, giving developers the opportunity to learn new ways to build things and implement APIs in their apps.
“At Apple, our goal is to put industry-leading technology directly into the hands of developers and create tools that enable them to build the best apps,” said Susan Prescott, vice president of Worldwide Developer Relations at Apple. “Agent coding greatly increases productivity and creativity, streamlining development workflows and freeing developers to focus on innovation.”
The release candidate for Xcode 26.3 is available to developers as of today and will likely be released in the next week or so.
