Cursor

Cursor is an AI-native code editor built as a fork of Visual Studio Code, designed to integrate large language models directly into the coding workflow. Founded in 2023 by Anysphere (Michael Truell, Sualeh Asif, Arvid Lunnemark, and Aman Sanger — MIT graduates), Cursor quickly became the most talked-about AI coding tool, raising $400M at a $2.5B valuation. It is used by engineers at companies including OpenAI, Shopify, Instacart, Midjourney, and Perplexity.

Cursor Tab: AI Autocomplete on Steroids

Cursor Tab goes far beyond traditional autocomplete. While GitHub Copilot predicts the next line, Cursor Tab predicts multi-line edits — it can suggest entire function implementations, refactors across multiple lines, and even anticipate your next edit based on the change you just made. It observes your editing patterns and proactively suggests the next logical change. For example, if you rename a variable in one place, Cursor Tab will suggest renaming it everywhere else. The completions are fast (typically under 300ms) and context-aware, drawing from your entire codebase rather than just the current file.

Cmd+K: Inline Code Generation and Editing

The Cmd+K shortcut (Ctrl+K on Windows/Linux) opens an inline prompt bar that lets you generate or edit code using natural language. Select a block of code and type "refactor this to use async/await" or "add error handling for network failures" — Cursor rewrites the selected code in place, showing you a diff of the changes before you accept. You can also use Cmd+K with no selection to generate new code at the cursor position. This is faster than switching to a chat panel because the AI operates directly in the editor context.

Codebase-Aware Chat

Cursor's chat panel (Cmd+L) is fundamentally different from ChatGPT or standalone AI assistants because it has deep awareness of your entire codebase. When you ask a question, Cursor automatically indexes your project files, understands import relationships, and retrieves relevant code context. You can ask "how does the authentication flow work in this project?" and Cursor will find the relevant files, trace the logic, and explain it — without you manually copying and pasting code into a chat window. You can also @-mention specific files, functions, or documentation to focus the AI's context.

Multi-File Editing with Composer

Composer (Cmd+I) is Cursor's most powerful feature for large changes. It can edit multiple files simultaneously based on a single natural language instruction. For example, you can type "add a new API endpoint for user preferences with the model, route, controller, and tests" and Composer will create or modify files across your project structure. It shows a plan of all changes before applying them, and you can accept or reject changes per file. This is transformative for refactoring tasks that touch dozens of files — work that would take hours manually can be completed in minutes.

.cursorrules: Project-Level AI Configuration

The .cursorrules file (placed in your project root) lets you define project-specific instructions for the AI. You can specify coding conventions ("always use single quotes," "use functional components, not class components"), architectural patterns ("follow the repository pattern for data access"), tech stack details ("this is a Next.js 14 project using App Router and Prisma"), and forbidden patterns ("never use any in TypeScript"). The AI reads these rules on every interaction, ensuring consistent output that matches your team's standards. This is especially valuable for teams where multiple developers use Cursor on the same codebase.

VS Code Foundation

Because Cursor is a fork of VS Code, it supports the VS Code extension ecosystem, keybindings, themes, and settings. Developers switching from VS Code can import their entire configuration — extensions, shortcuts, snippets — in one click. The editor looks and feels identical to VS Code, which eliminates the learning curve for the editor itself and lets developers focus solely on learning the AI features. Terminal, debugger, Git integration, and all core VS Code functionality remain intact.

Privacy and Context Control

Cursor offers a Privacy Mode that ensures none of your code is stored on their servers or used for model training. In Privacy Mode, code is sent to the AI model for processing but immediately discarded after the response is generated. Teams can also configure which files are indexed and which are excluded using .cursorignore (similar to .gitignore). Enterprise plans offer additional controls including SOC 2 compliance and the ability to use self-hosted models.