Developing two MVP products in 10 days: A comprehensive experience sharing with Cursor 2.0 – New agent-first UI, faster Composer 1 speed, multi-model quality optimization, built-in browser debugging, and cloud flexibility. 1. Brand new user interface (UI) Cursor 2.0 places the "agent" at the core, allowing users to focus on output results rather than underlying file management. This simplifies the development process and supports a "no-code" mode; if manual adjustments are needed, users can still switch back to the traditional IDE interface. 2. Composer 1 Model: While not as fast as models like GPT-5 Codex or Claude Sonnet 4.5 for complex tasks, Composer 1's response speed far surpasses its competitors. Suitable for rapid prototyping and simple feature development, it's an ideal choice for "Vibe Coding" and is widely used in amateur and client projects. 3. Parallel operation of multiple models allows the same prompt to be executed on multiple AI models simultaneously, selecting the best output. Tomar uses this to solve common pain points in UI generation (such as unsightly purple gradients). Although the cost is higher, it is of significant value to projects that pursue high-quality interfaces. 4. Built-in browser tools integrate real-time debugging functionality: Select an element, access developer tools, and instantly fix UI issues without switching applications. Drawing inspiration from Windsurf, this enhances development efficiency. 5. The built-in code review feature, "Agent Review," automatically scans for security vulnerabilities, potential bugs, and code quality, suitable for initial checks. Tomar suggests that for in-depth PR reviews, specialized tools such as @coderabbitai are still necessary. 6. Cloud-based Agents: Develop new features anytime, anywhere. Agents can be run remotely from devices such as mobile phones. Connect to a GitHub repository to add features, write blog posts, or handle marketing tasks. Tomar sees this as a highlight of its flexibility and is currently in the experimental stage.
