Apple has escalated its enforcement against a new category of AI-powered applications known as 'vibe coding' tools — apps that let users build other apps using natural language prompts and dynamic code generation. The crackdown began in early March when Apple quietly blocked updates to established developer tools including Replit and Vibecode, and intensified on 30 March when Apple completely removed the popular Anything app from the App Store.
The company is citing App Review Guideline 2.5.2, which states that apps must be 'self-contained in their bundles' and may not 'download, install, or execute code which introduces or changes features or functionality.' Apple also referenced Section 3.3.1(B) of the Developer Programme Licence, which restricts downloaded code from changing an app's primary purpose. The core issue is that vibe coding tools allow users to generate and run arbitrary code within the app — functionality that Apple considers a violation of its long-standing self-containment rules.
Anything developer Dhruv Amin attempted a workaround by submitting an update that would display app previews in a web browser rather than within the app itself. Apple rejected the submission and removed the app entirely. The app has since returned to the App Store as of 3 April, suggesting some resolution was reached, but the broader enforcement pattern continues to affect other vibe coding tools.
For context engineers, this story highlights a growing tension between AI-powered development tools and platform gatekeepers. The App Store's review process was designed for a world where apps were static bundles — not for a world where AI agents can generate entire applications on the fly. As vibe coding tools become more sophisticated and mainstream (global app submissions jumped 30% last year to nearly 600,000), Apple's response will shape whether mobile development remains a tightly controlled ecosystem or adapts to accommodate AI-native workflows.
