Cursor

Cursor is a commercial AI code editor and coding-agent platform. The desktop application combines repository search, chat, inline completion, multi-file editing, terminals, source control, and agent workflows in one editor. Official documentation also covers a command-line agent and cloud agents that can work from browser, phone, GitHub, Slack, Linear, and other integrations. The desktop download page currently provides builds for macOS, Windows, and Linux. Cursor does not publish a canonical open-source product repository or public software license for the editor.

Access is organized into live plans rather than one perpetual software license. The pricing page lists a free Hobby tier with limited agent requests and no credit card requirement, paid individual options, team plans, and custom enterprise arrangements. The free tier limits agent usage; paid plans add usage, collaboration, and administration features. Enterprise controls include repository, model, MCP, browser, network, audit, and identity-management options.

Top features

• Codebase-aware questions and search for understanding files, symbols, dependencies, and architecture.
• Inline completion and targeted editing for small changes without starting a longer agent task.
• Agent workflows that can plan, change multiple files, run commands, inspect results, and present diffs for review.
• Desktop editor downloads for macOS, Windows, and Linux.
• Terminal CLI and headless options for shell-driven work, scripts, and CI-oriented use.
• Cloud agents for delegated work outside the local editor, with integrations documented for common engineering systems.
• Customization through rules, skills, subagents, hooks, plugins, and MCP servers.
• Security and administration features including Privacy Mode and enterprise policy controls.

Use cases

• Understand an unfamiliar repository before choosing where to make a change.
• Implement and test a feature across frontend, backend, and test files while reviewing diffs in the editor.
• Use autocomplete for routine code and switch to an agent for debugging or larger refactors.
• Delegate a contained issue to a cloud agent and review the branch or pull request it produces.
• Standardize model, repository, network, and tool-access policies for a development organization.