Welcome to the Cursor IDE training exercises! These hands-on exercises will teach you the most powerful features of Cursor 2.0 through practical coding scenarios. Complete all exercises in approximately 63 minutes (or ~27 min for Foundations only).
By completing these exercises, you'll master:
- Intelligent Autocomplete - Let Cursor write code from comments and context
- Project Context - Use AGENTS.md to teach AI your project standards
- Coding Standards - Enforce patterns with auto-attaching Cursor Rules
- Context & Shortcuts - Navigate efficiently with @ references and keyboard shortcuts
- Safe Experimentation - Use checkpoints and duplicate for risk-free exploration
- Real-time Documentation - Get current library docs with MCP/Context7
- Agent Workflow - Use the Agents Window for autonomous coding
- Mode & Model Selection - Choose the right AI mode and model for each task
- Strategic Planning - Create executable plans with Plan Mode
- Parallel Productivity - Run multiple agents simultaneously for 3-4x speed boost
Before starting, ensure you have:
- ✅ Cursor IDE installed (latest version)
- ✅ Git configured on your system
- ✅ Programming knowledge in at least one of: C#, JavaScript, or Python
- ✅ Node.js installed (for Context7 MCP server - covered in Exercise 06)
Each exercise is language-agnostic in concept but includes starter code for:
- C# - Following m_ prefix conventions, #region blocks, XML documentation
- JavaScript - ES6+ patterns, async/await, modern module systems
- Python - PEP 8 conventions, type hints, pythonic patterns
Choose your preferred language and work through the exercises in order.
01. Autocomplete Magic (~5 min)
Learn: Tab completion, comment-to-code generation, commit message generation
What you'll do: Use Cursor's predictive capabilities to write methods from comments, navigate with Tab-Tab, and auto-generate commit messages.
02. AGENTS.md - The AI README (~5 min)
Learn: Project context documentation, narrative AI instructions
What you'll do: Create comprehensive AGENTS.md files that help AI understand your project holistically, including nested directory-specific context.
Learn: Auto-attach rules, glob patterns, enforced coding patterns
What you'll do: Create .cursor/rules that automatically enforce naming conventions, test patterns, and code quality standards when working in specific files.
04. Context & Keyboard Shortcuts (~3 min)
Learn: @ symbol references, essential keyboard shortcuts, explicit vs auto context
What you'll do: Master the @ symbol to add precise context (@files, @folders, @symbols, @URLs) and learn keyboard shortcuts (Cmd+I, Cmd+K, Cmd+L, Shift+Tab) for efficient workflows.
05. Checkpoints & Recovery (~3 min)
Learn: Automatic checkpoints, restore agent changes, Duplicate feature
What you'll do: Experience Cursor's safety net - use checkpoints to rollback agent changes, and Duplicate to fork conversations and explore alternative approaches.
06. MCP & Context7 (~4 min)
Learn: MCP server setup, Context7 real-time documentation, "use context7" prompts
What you'll do: Install the Context7 MCP server and use "use context7" to fetch current documentation, avoiding outdated code generation.
Fresh Start Recommended: Before starting Phase 2, reset exercise files to their original state:
git checkout -- exercises/cursor-exercises/exercises-files/
07. Basic Agent Workflow (~8 min)
Learn: Agents Window (Cmd+I), Review changes, Find Issues, agent execution
What you'll do: Experience the complete agent workflow - open Agents Window, create tasks, monitor progress, review changes, use AI-powered issue detection, and accept changes.
08. Agent Modes & Model Selection (~3 min)
Learn: Agent/Ask/Plan/Custom modes, Composer vs Claude Sonnet model selection
What you'll do: Switch between agent modes for different tasks, and choose the right AI model (Composer for speed, Sonnet for complex reasoning).
09. Plan Mode + Execute (~7 min)
Learn: Executable planning (Shift+Tab), plan critique, phase-by-phase execution
What you'll do: Use Plan Mode to create structured implementation plans for complex features, critique the plan to catch issues early, then execute phase-by-phase or all at once.
10. Multi-Agent Parallel Execution (~8 min)
Learn: Parallel agents, worktrees, 3-4x productivity boost
What you'll do: Launch multiple agents simultaneously working on independent tasks using isolated worktrees, achieving dramatic productivity gains through parallelization.
Learn: Multiple solution comparison, model selection, architectural decisions
What you'll do: Use Cycle Agent Count to generate multiple solutions from different models for a caching feature, then compare and select the best approach.
12. Browser Tool - Visual Debugging (~5 min)
Learn: Embedded browser, visual verification, E2E testing with AI
What you'll do: Use Cursor's browser tool to have an AI agent navigate websites, interact with forms, and verify visual elements - enabling AI-assisted end-to-end testing.
-
Choose Your Language
- Navigate to any exercise folder
- Open the language folder you're comfortable with (csharp/, javascript/, or python/)
- Each exercise README explains the concept in language-agnostic terms
-
Work Through Exercises in Order
- Exercises build on each other progressively
- Start with Exercise 01 and work sequentially
- Don't skip exercises - each introduces critical concepts
-
Read the Exercise README First
- Each exercise has detailed instructions
- Learning objectives are clearly stated
- Success criteria help you know when you're done
-
Use Cursor Features as Instructed
- Follow the prompts to practice specific Cursor features
- Experiment and observe Cursor's behavior
- Compare your results with the exercise goals
These exercises follow a learn-by-doing approach:
- No solution folders - You learn through practice and experimentation
- Real scenarios - Work with practical code problems, not toy examples
- Progressive complexity - Each exercise builds on previous knowledge
- Feature-focused - Master specific Cursor capabilities in each exercise
- Open relevant files in tabs - Cursor uses open tabs for better context
- Read exercise READMEs carefully - They contain important setup steps
- Don't rush - Take time to understand each feature
- Experiment - Try variations beyond the instructions
- Use keyboard shortcuts - Cmd+I (Agents Window), Shift+Tab (Plan Mode)
| Shortcut | Action |
|---|---|
| Cmd/Ctrl + I | Open Agents Window |
| Cmd/Ctrl + K | Inline edit selected code |
| Cmd/Ctrl + L | Open chat with current file |
| Shift + Tab | Enter Plan Mode |
| Cmd/Ctrl + T | New agent tab |
| Cmd/Ctrl + . | Cycle through agent modes |
Work through all 12 exercises at your own pace (~63 minutes total)
- New students: Complete all exercises (Phase 1 and Phase 2)
- Returning students familiar with basic Cursor: Skip to Phase 2 (Exercise 07+) for Core Agent features
- Time-constrained (~27 min): Focus on Exercises 01-06 (Foundations)
- Quick agent overview (~18 min): Exercises 07-09 only (Agent Workflow + Plan Mode)
After completing these exercises, you'll be ready to:
- Use Cursor 2.0 for real project work
- Implement the complete agent-centered workflow
- Leverage parallel agents for maximum productivity
- Create effective project documentation (AGENTS.md, rules)
- Plan and execute complex features confidently
- Navigate efficiently with keyboard shortcuts
- Use real-time documentation with Context7
- Choose the right AI mode and model for each task
- Cursor Documentation
- Full Workshop Guide - 84-minute comprehensive course
- Cursor Community
- Cursor Cheat Sheet - Quick reference for all features
If you encounter issues:
- Check the exercise README for troubleshooting tips
- Verify your Cursor version is up to date
- Ensure prerequisites are installed (Node.js for MCP)
- Review the full workshop guide for detailed explanations
- Refer to the cheat sheet for quick syntax reminders
Ready to start? Head to Exercise 01: Autocomplete and begin your Cursor mastery journey!