Cursor AI Complete Tutorial

Installation

Download from cursor.com. It imports VS Code settings automatically. All extensions, keybindings, and themes migrate instantly.

Three Interaction Modes

Tab (Inline Completion): Context-aware autocomplete within the current file.

Chat (Cmd+L): Conversational AI with codebase access. Ask questions, get explanations.

Composer (Cmd+I): Multi-file editing agent. Describe changes across your entire project.

.cursorrules Configuration

Create .cursorrules in project root to set AI behavior:

You are a TypeScript expert specializing in Next.js App Router.
Code Style:
Use TypeScript strict mode, never use any
Prefer const over let, never use var
Single quotes, 2-space indentation
Always add explicit return types
Component Conventions:
Functional components only
Props interface named ComponentNameProps
Server Components by default
Use cn() for className merging
Data Fetching:
React Query for client-side fetching
Never use useEffect for data fetching

Codebase Indexing

Cursor indexes your entire codebase. After indexing:

Use @codebase to query semantically across all files

Use @filename.ts to reference specific files

Use @src/folder/ to reference directories

Composer Examples

Refactoring:

Refactor Button component to support a ghost variant.
Update all usages in @src/components/

New feature:

Add dark mode toggle to header:
Use next-themes for persistence
Sun/moon icons from @/components/icons
Update ThemeProvider in @src/app/layout.tsx

Migration:

Migrate all files in @src/features/products/ from
direct fetch calls to React Query pattern in @src/features/users/api/queries.ts

Keyboard Shortcuts

ActionShortcut

ComposerCmd+I ChatCmd+L Accept suggestionTab Inline chatCmd+K Reject suggestionEscape

Pro Tips

Spend 1 hour perfecting your .cursorrules - it is the biggest multiplier

Use @codebase before making changes to understand existing patterns

In Composer, be specific: name files, describe current and target state

Use Cmd+K for small targeted changes, Composer for multi-file work