cursor-102-pro-prompting

Source: The Cursor 102: Pro Prompting Guide

• URL: https://www.youtube.com/watch?v=zGFroBGud7w
• Title: The Cursor 102: Pro Prompting Guide
• Channel: Indie Dev (Theo Browne)
• Key Topics: Cursor AI, Tooling, Prompt Engineering, Efficiency, Context Management

Summary

“Cursor 102” is a guide for advanced users to move beyond basic completions and use Cursor AI as a powerful development partner. It focuses on reducing iterations, managing context costs, and getting reliable code outputs.

High-Level Strategies (Minimize Turns & Costs):

1. Q&A Strategy: Don’t let AI guess. Ask it to “Ask me 2-3 clarifying questions” before writing code for complex tasks (e.g., Stripe integration, DB design).
2. Pros and Cons Strategy: Use AI to analyze trade-offs (e.g., Resend vs. Mailgun) before choosing a tool to implement.
3. Role Prompting: Assign specific roles like “Senior Security Engineer” for code reviews or “React Specialist” for refactoring.

Context & Efficiency Tips:

• Specific References: Use @file instead of @codebase whenever possible to save tokens and reduce noise.
• Rules for AI (.cursorrules): Move recurring style/architecture rules into a persistent file to save 300-800 tokens per prompt and increase consistency.
• Plan Mode: Demand a numbered plan and approval before the AI starts writing code (Shift+Tab in some versions or via explicit prompt).
• Concise Outputs: Explicitly tell AI to be brief and avoid “chit-chat” to reduce token burn (e.g., “max 4 sentences explanation”).
• Incremental Workflow: Break large features into smaller, testable steps (Skeleton → Auth → Logic) for easier debugging and cheaper model calls.