Cursor
Cursor
The editor most teams settled on, with an agent that ships whole features.
Updated
The read
Cursor is the default answer for a reason: completion that feels a step ahead of you, model choice you actually control, and an agent that has grown from editing one file to carrying a feature across a repository. It is at its best as the place you work all day, with the genuinely hard problems handed to something more deliberate. The free tier exists but is a taste. Real use means a subscription, and heavy agent work meters on top of it.
Where it shines
- Completion quality is still the benchmark others get measured against
- Model choice stays yours rather than the vendor's
- The agent handles multi-file work, not just the file you are sitting in
Worth knowing first
- Serious use needs the subscription; the free tier is really a trial
- Heavy agent use meters on top of what you already pay
Index score
81.8
Provisional. This score is read from the product's public capability, not from the one-prompt rebuild the fully reviewed entries went through, so treat it as a placing rather than a verdict.
- Rank
- 02 of 54
- Builds
- Tools
- Category
- AI code editor
- Output
- Edits and commits in your own repository
- Pricing
- Free tier
- Best for
- Daily driving, with an agent on call
Scores are our own editorial judgement, weighted craft 30, speed 25, control 25 and value 20.
Also worth a look
Full index- 03ConductorConductorProvisionalA free macOS app for running multiple Claude Code or Codex agents in parallel isolated workspaces.AI code editorFree tier81.8
- 07Devin DesktopDevin DesktopProvisionalThe editor formerly called Windsurf, now a control room for several agents.AI code editorFree tier80.5
- 08ZedZedProvisionalAn editor fast enough that the AI is never the thing you are waiting for.AI code editorFree tier80.4
