Windsurf and Cursor Rules

Rules

Modern development workflows benefit greatly from AI assistance. Without Windsurf, Cursor or other tools like ChatGPT, you're behind the competition, especially if you're a solopreneur.

You can find Windsurf and Cursor rules inside the project with the .windsurf file and .cursor folder.

• If using Windsurf, you can remove the .cursor folder
• If using Cursor, you can remove the .windsurf file

AI Development Tips

• Dedicated Feature Chats - Always implement one feature per chat to maintain context. When creating forms, keep all form development in one chat until perfect. Use separate chats for design work. This prevents context loss and keeps your workflow organized.
• Reference Existing Code - When implementing new features, always mention related files and patterns from your project. For subsequent similar components (like a second form), the AI will remember the context if you're working in the same chat.
• Inline vs. Cascade Edits - For small changes, use inline edits rather than cascade changes to save time. This focuses the modifications exactly where needed.
• Update Rules Regularly - Rules are crucial for consistent AI output. Keep updating them as your project evolves to ensure the AI understands your code patterns and preferences.
• Section-by-Section Design - Never request an entire landing page at once. Build designs section by section in the same chat for consistency, approving each piece before moving to the next.
• Functionality First, Design Later - For complex features, focus on functionality before aesthetics. Once the core works, switch to your design chat to improve appearance while maintaining your app's style.
• Break Down Complex Features - Divide complicated tasks into smaller, focused requests. This makes the development process more manageable and produces better results.
• Be Specific in Requests - Clear, detailed prompts with references to similar implementations yield better, more consistent code that fits your project architecture.
• Commit Frequently - When code is halfway complete or approaching perfection, commit what you have. This creates safety checkpoints throughout development.
• Choose Models Strategically - Don't default to thinking models for code tasks - they're time-consuming and often produce less compatible code. Best coding models are Claude Sonnet 3.7, GPT 4.1, and Gemini 2.5 Pro. For design work, Claude 3.7 excels, but experiment with other models too.
• Rules as Control Mechanism - Rules are your most powerful tool for controlling AI output. Craft comprehensive guidelines covering code style, architecture, state management, and design principles. Update these rules regularly as your project evolves to ensure consistent, production-ready code without extensive corrections.