Replit упрощает разработку мобильных приложений, позволяя запускать их в App Store без написания кода Translation: Replit simplifies mobile app development, allowing users to launch them in the App Store without coding knowledge

The AI startup Replit has launched its Mobile Apps feature, allowing users to develop applications using a method called vibe-coding and release them directly from the editor.

This new tool enables developers and entrepreneurs to «transform an idea into a functioning application in just minutes, and launch it on the App Store within a few days.»

The process involves:

To submit an application to the store, an Apple Developer account must be created.

Before appearing on the app store, the software undergoes a mandatory review process. Although Apple is known for its stringent requirements, the company claims to process the majority of applications within a day.

«Just tell Replit Agent what you want to create, repeat it in chat, instantly preview it on your phone, and publish when you’re ready,» the announcement states.

Replit’s platform is now integrating Stripe support for project monetization. The startup’s valuation in the upcoming funding round could reach $9 billion, up from $3 billion in September.

Vibe-coding is gaining traction as a rapid development technique that allows for app creation without extensive technical knowledge. However, this approach raises concerns among cybersecurity experts.

A recent study identified 69 vulnerabilities across 15 applications developed using popular tools such as Cursor, Claude Code, Codex, Replit, and Devin.

Tenzai specialists tested five AI agents on their ability to write secure code. In a controlled experiment, each agent was tasked with creating a series of identical applications, using the same prompts and technology stack.

Analyzing the results, analysts identified common behavior patterns and recurring failure modes. On a positive note, the agents effectively avoided certain types of errors.

None of the generated solutions exhibited SQL injection or XSS vulnerabilities, which have consistently ranked among the top issues in the OWASP rankings over the years.

However, the digital assistants showed poor results when dealing with complex architectures and allowed vulnerabilities in business logic.

«Human developers intuitively understand how workflows should operate. Agents lack this ‘common sense’ and mainly rely on instructions,» the study notes.

For instance, 4 out of 5 agents permitted attackers to create orders with negative amounts.

Cursor, Devin, and Replit made similar errors concerning pricing.

A specialized version of the GPT-5 programming lineup, known as GPT-5.2 Codex, has been made available in Cursor.

CEO Michael Truell reported that using OpenAI’s tool, the team developed a browser that ran continuously without issues for a week.

«It consists of over 3 million lines of code across thousands of files. The rendering engine was built from scratch in Rust, incorporating HTML parsing, CSS cascading, layout, text formatting, drawing, and a customizable JS VM,» Truell stated.

In terms of metrics, GPT-5.2 slightly outperformed Opus 4.5 on the SWE-Bench Pro.

It’s worth noting that in November 2025, Anthropic unveiled Claude Opus 4.5 — «the best programming model in the world, featuring agents and computer usage.»