Cursor

What is Cursor?

Cursor is an AI coding environment and agent platform for building and maintaining software. Its desktop editor combines codebase-aware chat, agentic editing, and predictive Tab completions. The broader product also includes a command-line agent, cloud agents that work in parallel on their own environments, mobile access, Slack task handoff, GitHub review, and scheduled or event-triggered automations.

Agents can explore a repository, propose a plan, edit multiple files, run commands and tests, and return changes for review. Secure codebase indexing and semantic search provide project context, while users can choose among supported frontier and Cursor models. MCP servers, skills, hooks, rules, and marketplace extensions customize the agent's behavior and tool access.

How to Use Cursor

1. Install Cursor or its CLI and open an existing repository.
2. Review privacy settings, repository trust, model choice, and tool permissions.
3. Ask a question or give the agent a scoped task with acceptance criteria.
4. Inspect its plan and restrict risky commands, credentials, deployment, and external actions.
5. Review the diff, run relevant tests, and verify generated behavior before committing.
6. Use cloud agents or automations only after defining triggers, budgets, branches, and review ownership.

Core Features

• Agentic development: Plans and implements multi-file tasks with terminal and test access.
• Codebase understanding: Uses indexing and semantic search to answer project-specific questions.
• Tab completion: Predicts code edits and continuations in the editor.
• Multiple surfaces: Runs in desktop, CLI, cloud, mobile, Slack, and GitHub workflows.
• Parallel cloud agents: Gives autonomous tasks separate environments for end-to-end work.
• Automations: Runs always-on agents from schedules or repository events.
• Customization: Supports rules, skills, hooks, MCP tools, models, and team marketplaces.
• Team governance: Adds centralized administration, privacy mode, SSO, analytics, and access controls.

Use Cases

• Feature implementation: Plan, edit, test, and explain a scoped change.
• Repository exploration: Find definitions, dependencies, and behavior across a large codebase.
• Debugging: Trace failures, inspect logs, propose a root cause, and prepare a tested fix.
• Code review: Summarize changes and flag potential defects in GitHub.
• Maintenance automation: Trigger agents for CI failures and repetitive repository tasks.

Pricing

Hobby is free with limited Agent requests and Tab completions. Individual Pro starts at $20 per month and extends Agent limits, model access, MCPs, skills, hooks, and cloud agents. Teams starts at $40 per user per month with administration, team marketplace, shared context, SSO, analytics, and privacy controls. Enterprise uses custom pricing and adds pooled usage, SCIM, audit logs, service accounts, and granular controls.

Frequently Asked Questions

Can Cursor work outside the desktop editor?

Yes. The current product includes CLI, cloud, mobile, Slack, GitHub, and automation surfaces.

Does Cursor automatically guarantee correct code?

No. Users remain responsible for reviewing diffs, running tests, checking security, and controlling deployment.

Is code used for training?

The pricing FAQ states that when Privacy Mode is enabled, code data is not used for training by Cursor or its model providers.