Apple Adds Claude and OpenAI Codex to Xcode for Agentic Coding

Apple has announced a major update to Xcode, its flagship developer tool, adding built-in support for AI agents that can take more direct control of the app-building workflow. The company says the shift is aimed at enabling “agentic coding,” an emerging and contested approach where AI systems don’t just suggest code, but act on a project with minimal human oversight.

The new release, Xcode 26.3, is available immediately as a release candidate. It integrates Anthropic’s Claude Agent and OpenAI’s Codex directly into Xcode.

According to details shared alongside the announcement, the integrated agents can autonomously:

• write code
• build projects
• run tests
• visually verify their own work

Apple framed the update as its most significant embrace of AI-assisted software development since it introduced intelligence features in Xcode 26 last year. The move also lands amid broader industry debate about “vibe coding,” a term used for delegating software creation to large language models.

Apple executive Tim Sneath described the change as moving beyond the limits of earlier AI features that primarily responded to a developer’s prompts without full project context or the ability to take independent steps.

“Integrating intelligence into the Xcode developer workflow is powerful, but the model itself still has a somewhat limited aperture,” Sneath said during a press conference Tuesday morning. “It answers questions based on what the developer provides, but it doesn’t have access to the full context of the project, and it’s not able today to take action on its own. And so that changes today.”

By embedding Claude Agent and Codex in Xcode 26.3, Apple is explicitly endorsing a more autonomous model of AI-assisted development one that can carry work from writing code through builds and tests, and even attempt visual checks of outcomes.