Cursor
The first AI editor we stopped second-guessing.
Most AI coding tools autocomplete a line and call it a day. Cursor reads the project. Open a repository and it builds context across files, so when you ask it to rename a concept or thread a new prop through three components, it follows the chain instead of guessing from the line under your cursor.
The day-to-day feel is the selling point. Tab completion is quick and picks up your habits fast, agent mode handles small features end to end, and the diff view keeps you in charge of what actually lands. Where it stumbles is ambition: hand it a sprawling refactor and it can charge off somewhere you did not ask for, so you learn to keep requests tight.
Pricing is the one place to read the fine print. The $20 Pro plan covers most people, but heavy use of premium models drains credits quicker than the marketing suggests. For anyone working in a serious codebase, it is still the tool we keep installed.
What we liked
- Reads the whole repository before it suggests an edit
- Multi-file changes that hold their thread across the codebase
- Tab completion is fast and learns your patterns within an hour
- You pick the model, including the frontier ones
What we did not
- The agent can over-reach on big refactors if you do not scope it
- Premium model usage burns credits faster than you expect
- Still a fork of VS Code, so it inherits a few of its quirks