Free lesson: audit your app’s accessibility with Xcode’s specialist agents

A free lesson from AI-Assisted iOS Development. The presale price is $375 until 6 November. Find out about the program →

Before you watch

This lesson comes from AI-Assisted iOS Development, our program on using AI at every stage of iOS development. It’s the last lesson of module 4, iOS Feature Implementation Using AI, taught by Ehab Amer.

By this point in the program, AI agents have built a new feature from a written plan, and it has been reviewed and merged with the help of Conductor. It looks finished. This lesson checks whether it’s accessible.

You don’t need anything from the earlier lessons to follow along. If you have Xcode 27 with an AI provider set up in its coding assistant, you can run the same two audits on your own app as soon as you’ve watched it.

Introduction

General-purpose review agents are good at checking an implementation against its plan, but they aren’t tuned for every concern. Accessibility in particular benefits from tools built specifically for the platform, and Xcode 27 ships with its own specialist agents from Apple to do exactly that.

In this lesson, you’ll use two of Xcode’s specialists — VoiceOver and Dynamic Type — to audit the merged implementation and identify any accessibility violations that slipped through the earlier review.

By the end of this lesson you’ll be able to:

• Identify which of Xcode’s built-in specialist agents applies to a given accessibility concern.
• Use the VoiceOver specialist to audit an implementation for screen-reader violations.
• Use the Dynamic Type specialist to audit an implementation for text-scaling violations.
• Turn the findings from both specialists into concrete action items.

Demo: Auditing with Xcode’s Specialist Agents

Use Xcode’s built-in VoiceOver and Dynamic Type specialist agents to audit the implemented feature, and turn their findings into action items.

Try it yourself

To follow along exactly, download the project as it is at the start of this lesson.

Each audit in the video takes a single prompt. Open that project, or one of your own, in Xcode 27, start two conversations, and paste one prompt into each:

/accessibility-voiceover-specialist review the implementation of the application and write the audit violations in a dedicated file.

/accessibility-dynamic-type-specialist Audit the implementation of this application, write all the violations you find in a document.

Then do what Ehab does at the end of the video: read both reports yourself, pick out the false positives, and decide which findings are worth fixing.

If you want to see how Apple’s specialists work, they’re plain markdown files. You’ll find them in your user folder under Library ▸ Developer ▸ Xcode ▸ CodingAssistant ▸ codex ▸ skills ▸ __xcode.

Conclusion

In this lesson, you closed the review of your AI-implemented feature with Xcode’s built-in specialist agents. The VoiceOver specialist audited how the feature behaves for screen-reader users, and the Dynamic Type specialist checked that its layout holds up as text scales, giving you a platform-aware pass that a general-purpose reviewer couldn’t match.

The key takeaway is that accessibility deserves tools built for the platform. Conductor’s parallel review is good at checking an implementation against a plan’s stated goals, but catching VoiceOver and Dynamic Type violations is a job better suited to agents that understand Apple’s accessibility APIs directly.

In the full program, this lesson completes a module that carries a feature from a written plan all the way through implementation, parallel review with Conductor, and a specialist accessibility audit. The next module builds directly on this work, using these audits to write unit tests for the feature.

Where this lesson fits

AI-Assisted iOS Development works through every stage of iOS development with AI, from exploring a legacy codebase to shipping on the App Store. It’s what we found worth adopting, and how to do it.

• Get started. Set up AI tooling in Xcode, Claude Code and Cursor, and use it to understand and modernise a legacy codebase.
• Plan and build. Shape a feature with AI, then implement it with agents working in parallel. This lesson ends that stage.
• Validate. Turn audits like these into tests, and review AI-written code so its mistakes don’t repeat.
• Ship. Add agentic code review to your CI pipeline, and use AI for App Store copy and user feedback.

Five of the eight modules are out now, and the rest arrive by mid-November.

Get the program

It’s $375 until 6 November, and $500 after that. You get all eight modules for life, homeworks with answer keys, quizzes, office hours with the Kodeco team, and a private Discord channel.

Get the program for $375 →

Already a Kodeco subscriber? The program’s content is part of your subscription. You can add the homeworks, quizzes, office hours and Discord channel for $70. See your options →

Not ready yet? Read why we built it, or see the full syllabus.