Cursor, built by Anysphere, is a fork of VS Code with AI woven into every layer of the editing experience. It went from curiosity to default editor for a huge slice of professional developers in under two years. Here’s what it’s actually like to use day to day.
What Makes Cursor Different
Most AI coding tools are plugins. Cursor is a fork - which sounds like a liability (you’re betting on Anysphere keeping pace with upstream VS Code) but is exactly what enables its best features. Because it owns the editor, Cursor can do things an extension never could: inline diffs you accept with a keystroke, predictions that edit code around your cursor, and an agent that operates the editor itself.
The Three Modes
1. Tab (Autocomplete)
Cursor’s autocomplete is the feature that hooks people. Instead of completing the current line, it predicts your next edit - sometimes several edits away, in a different part of the file. Change a function signature, and Tab will offer to update the call sites. It feels less like autocomplete and more like the editor reading your mind.
2. Chat / Inline Edit
Ctrl+K for inline edits, Ctrl+L for chat. Both are context-aware: Cursor indexes your codebase so answers reference your code, not generic examples. The indexing happens locally and is fast even on large repos.
3. Agent Mode
Agent mode is Cursor’s answer to Claude Code and friends. Give it a task - “add pagination to the users endpoint and update the tests” - and it plans, edits multiple files, runs terminal commands, and iterates. It’s competent, and it’s improving rapidly, though dedicated CLI agents still edge it out on very long-horizon tasks.
Codebase Indexing
Cursor builds a semantic index of your repo, so @-mentioning a file or asking “where is authentication handled?” returns real answers. This is the unglamorous feature that makes everything else work - without it, the AI is just a chatbot with an API key.
Pricing Reality Check
Cursor’s pricing has tiers (Hobby, Pro, and up), and the model-usage economics have been a moving target. The practical advice:
- Pro is worth it if you code daily - the Tab completions alone justify it.
- Watch usage-based limits on premium models; heavy Agent-mode users can burn through included usage.
- Bring-your-own-key is an option if you want predictable costs with specific providers.
Weaknesses
- Fork drift. VS Code upstream features and some extensions occasionally lag or misbehave.
- Resource usage. It’s Electron plus an indexer plus AI requests; on big monorepos it’s not light.
- Agent reliability varies. Great for well-scoped tasks, wandery for vague ones.
Verdict
Cursor earned its hype. If you want AI that lives in your editor rather than next to it, it’s the benchmark - see our Cursor vs GitHub Copilot comparison for how it stacks up against the incumbent, and browse alternatives to Cursor if a fork of VS Code isn’t your thing.