Frequently Asked Questions About Cursor

The most important answers about Cursor

Cursor is an AI-powered development environment that lets you write code, explore existing projects, and have agents implement changes. You can start with free access, but model usage, limits, and privacy settings vary by subscription and the features you use.

What is Cursor and what is it used for?

What is Cursor?

Cursor is an AI-powered development environment for writing, understanding, and modifying software. You can receive code suggestions as you type, ask questions about files, and have an agent perform programming tasks. Cursor thus combines the functionality of a code editor with AI assistance that can work within your project.

What can you do with Cursor?

You can have features suggested, error messages investigated, tests prepared, and existing code explained. Agents can also modify multiple files and perform available checks. How useful the result is depends on your assignment, project context, and evaluation. More generated code does not automatically mean a better solution.

Is Cursor an editor or an AI model?

Cursor is the environment in which you work. The AI responses and suggestions come from available models. This distinction is important: model selection affects capabilities, speed, and costs, while the editor provides access to files, the terminal, and other tools. Furthermore, a model may perform differently depending on the type of task.

Wat is het verschil tussen Cursor en Visual Studio Code?

Cursor builds on VS Code and adds its own AI features and agent control. As a result, many elements feel familiar. However, they are separate applications with differences in support and extensions. Check your most important extensions and workflows before moving an entire project to a different environment.

What is the difference between Cursor and GitHub Copilot?

Cursor offers its own development environment with integrated AI control. GitHub Copilot is available within various editors and GitHub workflows. Both can assist with code and agents. Compare this to your daily workflow: keep your existing editor, use project context, review changes, and manage costs. Neither takes technical control away from you.

Who is Cursor suitable for?

For developers who regularly work on existing or brand new software and want to use AI within their programming environment. Beginners can ask for explanations when encountering unfamiliar code. Experienced users can speed up repetitive tasks. Cursor is especially useful when you can define tasks and evaluate suggestions independently.

Can you program with Cursor without any experience?

You can quickly get an initial proposal, but you’ll need expertise to assess functionality and security. That’s why you should also use Cursor as a learning tool. Ask what assumptions a solution is based on and how you can verify them. Just because an application starts up doesn’t mean it’s automatically suitable for real users.

Install Cursor and switch

How do you get started with Cursor?

Download Cursor from the official website, install the correct version, and sign in. Then open a small project folder where you can practice safely. First, ask for an explanation of an existing feature. After that, you can make a minor change and see how the review of modifications works.

Werkt Cursor op Windows, Mac en Linux?

Cursor offers downloads for these desktop platforms. Choose the version that matches your operating system and processor. Check the current platform requirements if you’re using an older computer. Your programming language, local development server, and test environment may also have additional requirements for memory and storage space.

Do you need a powerful computer?

For online AI processing, you don’t need to run a large language model yourself. However, your computer must be able to handle your editor and project. Large projects, containers, and extensive testing can require significant processing power. AI assistance does not change the local requirements of your build or development environment.

Can you open existing projects?

Yes. You can open an existing project folder and continue working with the files it contains. First, check whether the project runs smoothly without any AI modifications. This provides a clear starting point. If something goes wrong afterward, you’ll be better able to determine whether the problem already existed or was caused by a modification.

Can you import settings from VS Code?

Yes. Cursor offers an import feature for settings, themes, keyboard shortcuts, and extensions, among other things. After importing, check your most important components. Importing a setting does not mean that every extension will work identically. For example, test formatting, debugging, and language support before you start important work.

Do all VS Code extensions work in Cursor?

No, full compatibility is not guaranteed. Cursor uses the Open VSX registry and its own distribution channel for extensions. Many well-known extensions are available, but not every VS Code extension is included or functions the same way. Be sure to check the extensions that your project or daily workflow depends on.

Can you use Cursor alongside VS Code?

Yes. They are separate applications, and you can open the same projects with them. Be mindful of simultaneous changes to the same files. Save your work and use version control so you can see what has changed. Also, check whether both environments apply the same formatting rules.

Does your project need to be on GitHub?

You don’t need to put every project on GitHub to edit local files. Cloud agents and certain integrations have their own requirements for repository access. So, make a distinction between local editor functions and tasks that run in an external environment. Public publishing is not a general requirement for using Cursor.

Cursor Tab and asking questions about codes

What is Cursor Tab?

Cursor Tab is the feature that suggests possible code changes while you’re editing. These can include additional lines, imports, or related adjustments. You decide which suggestions to accept. Be sure to read the suggestion in context; a logical addition to the current line might still cause a problem elsewhere.

How do you accept a Tab suggestion?

With the default settings, you accept a complete suggestion using the Tab key. You can customize this assignment via the keyboard shortcut settings. If you notice unusual behavior, also check whether an extension is using the same key. Do not accept a major proposal until you have reviewed which rules are actually changing.

Can Cursor explain existing code?

Yes. Select the relevant code or specify a file and request an explanation. For example, ask for a description of where data comes from and which other functions are called. Then verify the explanation against the project. A model may convincingly but incorrectly interpret unknown components.

Can you contact Cursor in Dutch?

You may use Dutch questions and commands. Leave technical names, code, and error messages as they are. If necessary, specify that you want explanations in Dutch, while code comments should follow the existing project language. This prevents a local preference from unintentionally leading to multiple languages in the source code.

Which programming languages can you work with?

Cursor can be used with a variety of programming languages. The quality of AI assistance depends on the model and the available context. Support for syntax highlighting, debugging, and formatting also comes from the editor and extensions. Therefore, test your entire workflow, not just a single successful AI response.

Can Cursor resolve error messages?

Cursor can investigate causes and suggest changes. Please provide the relevant error message, the expected behavior, and the versions used. If anything is unclear, ask for a diagnosis first. A quick fix may make an error message disappear without actually resolving the underlying problem.

Can Cursor write reviews?

Yes. You can have tests prepared using the framework your project is already using. Describe the desired behavior and important exceptions. Verify that the tests would actually reject an incorrect implementation. Simply confirming that the brand new code follows its own assumptions offers little certainty.

Can Cursor update documentation?

You can have explanations, installation instructions, and examples adapted to reflect any changes. Ensure the documentation aligns with verified behavior. Be sure to review commands, configuration settings, and version numbers. A readable manual that isn’t actionable will create extra work for users and colleagues down the line.

Working with Cursor Agent

What does Cursor Agent do?

The agent can examine project files, make changes, and use terminal commands. This allows it to tackle a task in multiple steps. The offers available depend on your settings and the tools at your disposal. Clearly describe the goal and review the changes before marking them as complete.

Can an agent edit multiple files?

Yes. This is useful when a change affects functions, calls, and tests. Review the files together. Also check whether any unnecessary changes outside the scope of the assignment have been made. A minor functional adjustment should not inadvertently introduce a brand new dependency or require extensive restructuring.

Can Cursor execute terminal commands?

Yes, agents can use the terminal when permitted by the settings. Examples include tests, builds, and development commands. Check which execution mode is active. A command can do more than just check code; for example, it can modify files or connect to an external service.

Do you have to approve every agent action?

That depends on the execution mode and permissions. Cursor has settings that determine which offers proceed automatically and when user review is required. Choose what works best for your project. Understand the effect of a setting before granting an agent broader access.

What is the difference between a local agent and a cloud agent?

A local agent works with the files and resources available on your local platform. A cloud agent performs tasks in a remote environment with its own access and configuration. When using the cloud, verify how repository access, required settings, and results are managed. Not every local dependency is automatically available there.

What tasks are best delegated to an agent?

Choose tasks with clear boundaries and a verifiable outcome. For example, adding a missing validation or fixing a described error. Specify what needs to be preserved and what checks are available. A request like “make everything more modern” leaves a lot of room for choices you might not intend.

Example of a specific request

Investigate why the contact form accepts an empty email address. Use the existing validation and test setup. Fix only this behavior; keep the field names and do not change the product type. Add an appropriate test and specify which checks you have performed.

How do you check what Agent has changed?

Review the change log and read the differences for each file. Check not only the added code but also the removed lines. Compare the results with your genuine assignment. Then perform the appropriate checks. A summary from the agent is helpful, but it does not replace an assessment of the actual changes.

Can you undo agent changes?

Cursor heeft checkpoints waarmee je binnen ondersteunde workflows naar een eerder punt kunt teruggaan. Die worden lokaal bewaard en staan los van Git. Gebruik daarnaast versiebeheer voor duurzame projectgeschiedenis. Controleer na herstel of ook je eigen tussentijdse wijzigingen nog aanwezig zijn.

Can Cursor build a complete application on its own?

You can seek support for large-scale development, but a broad project requires many decisions regarding behavior, data, and maintenance. It’s better to work in small, manageable increments. An initial demonstration is not the same as an application that reliably handles real accounts, payments, and user data.

Project context, rules, and better results

How does Cursor understand your project?

Cursor can use available files and relevant context to answer questions and perform tasks. An agent can search for relevant code on their own. That doesn’t mean they’ve reviewed every single part. When important conclusions are drawn, ask which files and assumptions they’re based on.

Do you need to attach all files to a question?

No. Start with the problem, relevant files, and the desired outcome. Add information only when it’s missing. A large amount of unrelated context doesn’t automatically make a task any clearer. For troubleshooting, a specific error message and a reproducible example are often more valuable than a lot of unrelated files.

What are Cursor Rules?

Rules are reusable instructions for Agent. You can use them to define project conventions, such as naming conventions, architecture, and test commands. Keep them specific and up to date. A rule like “write good code” provides little guidance; a reference to the existing validation pattern is easier to verify.

Where can I find project rules?

Project rules can be defined in .cursor/rules as supported .mdc files. Cursor also supports agent instructions via AGENTS.md. Check the scope of a rule. Some instructions always apply, while others apply only to specific files or when the rule is explicitly used.

Do the Rules also apply to Cursor Tab?

No. The documentation distinguishes between instructions for Agent and other AI features. Rules do not control Cursor Tab in the same way. Therefore, do not expect a brand new project rule to automatically change every autocomplete suggestion. Check the support for the specific feature.

How can you write a better assignment?

Describe the desired outcome, the current situation, and your constraints. Include examples when behavior can be interpreted in multiple ways. Also specify which parts fall outside the scope of the assignment. This allows Cursor to work more effectively and makes it easier for you to assess whether the solution is correct.

Why does Cursor sometimes invent features or settings?

Een AI-model kan aannemelijke patronen combineren die in jouw project niet bestaan. Ook kan het uitgaan van een andere bibliotheekversie. Controleer voorgestelde API’s in de werkelijke afhankelijkheden en officiële documentatie. Geef gevonden beperkingen terug, zodat een vervolgvoorstel op de juiste informatie kan steunen.

How do you prevent a conversation from going off track?

Stop when the premise is no longer valid and identify which assumption is incorrect. If necessary, start a brand new conversation with a brief, accurate summary. Adding more and more explanation to a derailed approach doesn’t always help. Save useful changes first, so you can retain them in a targeted manner.

Free use, subscriptions, and costs

Is Cursor free?

Free access with limited AI usage is available. This allows you to explore the available features. Check the current limits in your account. Older plans with fixed request limits may no longer match the subscription currently being offered.

What Cursor subscriptions are available?

The current pricing page lists Free, Plus, Premium, Super, and Ultra, with separate team and enterprise options. In addition, older subscriptions are available for existing customers. Be sure to compare the terms and conditions associated with your account. The same product name in an older manual may have different terms of use.

How much does Cursor cost?

The current pricing page lists, among other things, $8 per month for Plus, $30 for Premium, $100 for Super, and $200 for Ultra. Taxes and certain model-specific costs may apply. Please check the total price and the included storage space before purchasing.

Why do you also see prices for Pro and Pro+?

Pro and Pro+ are part of previous Cursor subscriptions that can continue for existing customers. Their usage pools and additional costs work differently from the newly offered plans. Therefore, please refer to your own billing dashboard and the accompanying documentation, rather than combining rates from different subscription types.

Are all AI models included in the monthly fee?

No. Under the brand new subscription model, a distinction is made between included usage and separately paid usage of external models. Check model access and usage separately. Just because you can select a model doesn’t automatically mean it’s covered by your fixed monthly fee without additional costs.

Can Cursor incur additional costs?

Yes, depending on your subscription, model selection, and any additional usage options you’ve configured. Review the pricing settings before running a large number of tasks. An agent may require multiple model interactions during a single task. Therefore, base your estimate on your actual workflow and not just on the number of messages you send yourself.

Which subscription plan suits your needs?

First, test representative tasks and assess resource usage. Someone who occasionally asks for guidance uses Cursor differently than someone who performs multiple agent tasks daily. For teams, centralized management and data agreements are also important factors. Only upgrade to more capacity once you know where your current limitations lie.

Privacy, confidential codes, and rights

What is Privacy Mode?

According to Cursor, Privacy Mode prevents your code from being used for training by Cursor and the relevant model providers. You can check this setting in Cursor Settings. For teams, an administrator can enforce this setting centrally. Always verify the setting of the environment in which you are actually working.

Does Privacy Mode mean that everything stays local?

No. For AI features, commands and relevant code context may be sent to Cursor and model providers. Privacy Mode addresses, among other things, training usage and data protection. It is not a general guarantee that all processing takes place on your computer or that every feature follows the same storage rules.

Can you exclude files from the AI context?

With a .cursorignore file, you can set additional exclusions. Carefully review patterns and ranges. Important: Terminal commands and MCP tools may bypass these file filters. Therefore, consider an exclusion file as a tool to restrict access, not as complete protection for secrets.

Can you use Cursor with a confidential customer code?

First, review customer agreements, organizational policies, and the terms and conditions for your account. Consider model providers, processing, and access rights. Do not use real secrets in examples. Privacy Mode is relevant, but it does not address every question regarding confidentiality, storage, and the use of external tools.

Do the same privacy rules apply to custom API keys?

Not automatically. Cursor notes that the zero-retention agreements do not apply in the same way when you use your own API keys. In that case, the terms and conditions of your model provider also come into play. Assess that combination before processing confidential material. A different payment method may also entail different data agreements.

Can you use your code from Cursor for commercial purposes?

Cursor states that customers own the generated code. Additionally, verify any rights to existing code, dependencies, and incorporated components. An AI suggestion is not proof of exclusivity or accuracy. For client work, your contract, project keys, and quality agreements also remain important.

Troubleshooting and Making Changes

Why aren’t you getting Tab suggestions?

Check that the feature is enabled, that you’re logged in, and that there are no usage restrictions. Also check for conflicting extensions or keyboard shortcuts. Test a simple file in a small project. This will help you distinguish between a general issue and behavior that occurs only in a specific environment.

What should you do if the cursor slows down?

Check whether the issue is related to a large project, numerous extensions, or a specific agent task. Close unnecessary processes and test where necessary without any suspicious extensions. Keep your installation up to date. When submitting a support request, note exactly what is running slowly: typing, searching, AI responses, or local checks.

What should you do if an agent modifies the wrong files?

Pause the task and review the change log. Save useful changes and selectively revert unwanted ones. Then issue a smaller task with explicit boundaries. Also check whether project rules contain outdated or conflicting instructions. Greater autonomy doesn’t help if the starting point is wrong.

What should you check before publishing changes?

Review the changes, perform the appropriate checks, and compare the results with the requirements. Also verify what the agent did not test. For example, a successful build says little about incorrect access permissions. Evaluate the behavior from the user’s perspective and in light of existing project agreements.

• Functionality: Does the result work correctly for both standard and non-standard input?
• Scope: Have only the necessary files and dependencies been modified?
• Security: Are data protection and access control maintained?
• Maintenance: Have relevant tests, documentation, and version agreements been verified?

Which cursor function is best suited for which task?

Key insights about Cursor

• Cursor combines an editor with AI assistance: you can autocomplete text, ask questions, and have agents perform tasks.
• Check the migration process from VS Code: settings are transferable, but extensions may not always work the same way.
• Clear boundaries make changes easier to evaluate: specify the goal, context, and acceptance criteria.
• Compare subscription and product costs side by side: included usage differs from separately paid product usage.
• Privacy and code verification remain essential: check data settings and test changes before publishing.