As is well known, the spirits differ through the use of artificial intelligence in development: For many, AI has long been indispensable in everyday life in programming as a helper. But leave the complete development of an app to a AI? This is still a red cloth for a larger number of programmers. The developer Indragie Karunaratne wants to break a lance for the AI and has published an example with the Mac app Context, which is now possible.
Context is an app for debugging MCP servers. The Model Context Protocol (MCP) is a standard of Anthropic for the connection between AI applications and external data sources. The app was developed in Apple’s programming language Swift. The user interfaces were written in Swiftui. The Source text of the app Can be publicly viewed in the developer’s Github account. Karunaratne works full -time as a technology director at Sentry, a platform to determine and remedy mistakes in applications.
A dream for leisure projects
In one Detailed blog post The experienced MacOS developer shows that the app was created almost entirely with Anthropics Claude Code. He wrote less than 1,000 of the 20,000 code lines-the rest comes from the AI-based development environment.
The developer, who has been programming software for the Mac since 2008, describes his experience as fundamentally different from previous projects. While he used to be able to complete a side projects in his free time for years, the AI enabled him to develop and publish a complete MacOS application within a few months.
Strengths and weaknesses of the AI
Claude Code differs significantly from other AI coding tools such as Github Copilot, Cursor or Windsurf. Instead of expanding existing IDEs with AI functions, it replaces the traditional development environment entirely with a terminal-based surface. The focus is on an “agent” approach.
The development of Context also became aware of the strengths and weaknesses of the AI. Claude code mastered Swift features to version 5.5 good, but have difficulties with Swift Concurrency, which was introduced from Swift 5.5. Often the system falls back on outdated objective-C-APIs instead of using modern Swift alternatives.
At Swiftui, however, Claude Code showed itself from its better side: The generated user interfaces worked reliably, even if they look not very appealing at first. It is noteworthy that the design was significantly improved with simple instructions such as “Make It More Beautiful”, as Karunaratne writes.
Preparation is everything
Another knowledge from the project was that “prompt engineering” is no longer a big challenge today, since modern models also understand incomplete and poorly formulated inquiries. The key to success, on the other hand, was the “Context Engineering”. The limitation to 200,000 tokens among the latest Anthropic models was a challenge.
Despite all AI support, precise specifications are essential. The developer warns of marketing demos who claim to be able to create complete apps with a single set. For production -ready software, detailed descriptions of the desired functionality are still necessary. Another tip: Claude Code should first create a plan with “Ultrathink” before it begins with the implementation. This instruction activates an expanded thinking process that provides better quality results.
Set up feedback loops
Claude Code is particularly effective through well-equipped feedback loops. The system should be able to compile, test and fix errors independently. The developer used additional tools such as XcodebuildMCP to facilitate Claude Code to create MacOS apps. Screenshots can be inserted directly into Claude Code for troubleshooting. The system then analyzes visual problems and suggests corrections.
Finally, Claude Code created a 2,000 lines long Python script that automated the release process, it automatically signed the code, notarized it, created DMG files and distributed updates.
Big changes ahead?
The developer predicts a great change in the development environment. Traditional features such as file rooms or source code editors in development tools could become more important. Instead, Ides will concentrate on helping developers with the priming of AI agents and setting up feedback loops. Classic programming could become a secondary activity.
Discover more from Apple News
Subscribe to get the latest posts sent to your email.