AI tool directory and guides
Features · Pricing · Tutorials

Cursor

Cursor is an AI coding agent for planning, editing, testing and reviewing software.

Cursor illustration
FINDGOODAI / USE CASE NOTES
Cursor: Use cases
1. Implement a feature from project requirements.
2. Investigate an error and work through a fix.
3. Understand an unfamiliar repository.

About Cursor

Cursor works in the editor, terminal and cloud. Agents use project context to make changes, run tools and prepare results for review. Developers can choose models, connect tools through MCP, and delegate work to cloud agents.

Main features

  • Explore a codebase, plan a change and edit project files.
  • Run development tools and tests with Agent.
  • Use Tab completions and focused code edits in the editor.
  • Delegate tasks to cloud agents.
  • Review GitHub pull requests with Bugbot.
  • Connect external tools through MCP, skills and hooks.

Inputs, outputs and access

ItemDetails
Task inputText instructions and project source files.
Task resultsFile changes, explanations, command and test results, and review findings.
Keeping the resultChanges stay in project source files and can enter Git or pull-request workflows.
PlatformsmacOS, Windows and Linux desktop apps; CLI, web cloud agents and iOS.
Service connectionAI requests use Cursor services; cloud agents execute remotely.
Workflow extensionsMCP tools, skills, hooks and GitHub pull-request review.
AccountAn account is required for service features covered by account access.

Capabilities in practice

Work from a task to reviewable project changes

Open the repository and describe the intended change with its project context. Agent can inspect files, plan the work, edit code and run development tools or tests. Tab completion and focused edits support smaller changes, while cloud agents take delegated tasks. The result remains a set of project changes that can be inspected and taken through the team’s Git review process.

Follow a change after deployment with Rollouts

The September 23, 2026 update adds Rollouts for Teams and Enterprise. It turns a pull-request diff into an editable monitoring plan, then checks deployment signals, logs, metrics and traces separately for each environment. Suspected regressions can lead to a proposed revert PR or a cloud-agent fix, with merging and rollback still left for review.

Check exploitable issues in pull requests

Security Review follows input through the codebase to find issues such as injection, authorization bypasses, exposed secrets and unsafe dependency changes. A finding includes severity, an attack path and a proposed fix. Teams can add repository-specific rules and enable the reviewer for selected repositories; draft PRs are skipped, while general style and quality review remains with Bugbot.

See the product in use

Review an MCP tool before execution
Review an MCP tool before execution · The tool call shows its name and parameters before approval.
Inspect the MCP tool result
Inspect the MCP tool result · The response remains visible with the tool parameters and returned data.

Platforms and access

Desktop for macOS, Windows and Linux, plus CLI, web cloud agents and an iOS app.

Pricing and plans

Monthly USD prices: Pro USD 20, Pro+ USD 60, Ultra USD 200; Teams Standard USD 40/user/month and Premium USD 120/user/month. Enterprise is custom priced. These USD prices exclude applicable taxes. The India-only Start plan is INR 649/month, tax inclusive. Usage pools reset monthly; model choice affects consumption, and on-demand usage can add cost.

Open source and deployment

Cursor is distributed under its service license.

Data and privacy

Inputs and suggestions are not used for model training except security review, feedback you submit, or explicit agreement. Privacy Mode adds model-provider controls; AI requests still use Cursor’s backend.

Frequently asked questions

Can I try Cursor for free?

Yes. Hobby includes core features with limited Agent, Chat and Tab usage.

Does Pro mean unlimited use of every model?

No. Included model usage depends on the pool and model; additional usage may be billed separately.

Can Cursor work on my existing project?

Yes. Open the repository and give Agent a task; it can inspect files, edit code and run project tools.

What does Privacy Mode change?

With Privacy Mode enabled, your data is not used for training, and technical and contractual protections apply to model providers.

Does it work outside the desktop editor?

Yes. You can use CLI, web cloud agents and iOS; feature access depends on your plan.

Product updates

Product and provider

Suitable tasks and users

Implement a feature from project requirements. Investigate an error and work through a fix. Understand an unfamiliar repository. Review a pull request before merging. Developers building and maintaining software. Independent builders working on prototypes and products. Engineering teams reviewing and coordinating code changes.

Availability

Available for use.

Ways to work

Use Cursor in its desktop editor, terminal CLI, web cloud agents and iOS app.

Support contacts

Support: hi@cursor.com; vulnerability reports: security-reports@cursor.com.

Category

Coding

Related tasks

Software development and code review.

Topics

Coding agent, repositories and code review.

Delivery

Local client with cloud AI and hosted cloud agents.

AI tasks

Coding, review and developer-tool execution.

Interaction

Agent chat, inline completion, terminal and pull requests.

Download

Download desktop clients through the Cursor download page.

Before starting

Provide the project repository and its development tools, with the required file and execution permissions.

Value proposition

Cursor is an AI coding agent for planning, editing, testing and reviewing software.

Access and help

Legal urlsService terms, privacy and applicable organization agreements.

Files and results

Input formats

Project source files and text instructions.

Editing results

Review file differences and edit the generated project source in your development environment.

From a task to a result

Start with a concrete task

Download Cursor for your platform and open a project. → Describe the task; give Agent the relevant files, constraints and desired result. → Review the changes and test output before committing or opening a pull request.

Platforms and availability

Platforms and limits

Desktop for macOS, Windows and Linux, plus CLI, web cloud agents and an iOS app. Hobby has limited usage; paid plans also have model-dependent included usage. Additional model usage can incur charges after the included pool is consumed. Cloud agents and connected tools need their own access and environment configuration.

Platforms

Windows, macOS, Linux, web and iOS.

Documentation language

English product documentation is available.

Account access

An account is required for subscribed AI features.

Network

AI, marketplace and updates communicate with Cursor backend services.

Subscriptions and billing

Access model

Free access conditions, subscriptions and additional usage are described in the plan row.

HobbyUSD 0 · Individual; limited usage. Limited Agent, Chat and Tab usage.
ProUSD 20/month · Individual. Cursor Models and Other Models pools.
Pro+USD 60/month · Individual. Cursor Models and Other Models pools.
UltraUSD 200/month · Individual. Cursor Models and Other Models pools.
Teams StandardUSD 40/month · Per user.
Teams PremiumUSD 120/month · Per user.
StartINR 649/month · India only; Cursor Models pool; no Other Models pool.
EnterpriseContract pricing · Custom quote.

Apps, integrations and development

Launch and manage cloud work

Cloud Agents API v1 is in public beta. It creates a durable agent, then schedules individual runs for prompts against a repository.

Webhook version boundary

Webhooks are not yet available in v1. Existing webhook integrations must use the legacy v0 surface until v1 support is released.

API rate limits

The default API limit is 20 requests per minute unless an endpoint defines another limit. Team, analytics and organization APIs have endpoint-specific allowances.

Bring project tools into Agent

Cursor connects to external MCP servers using local stdio, SSE or Streamable HTTP. It can consume tools, prompts, resources and interactive MCP app views.

Per-project and personal connections

Configure project tools in .cursor/mcp.json or shared user tools in ~/.cursor/mcp.json. Remote servers can use OAuth; local commands use configured environment credentials.

Protocols

REST endpoints and streaming run events.

Authentication

Basic or Bearer API-key authentication, including service-account keys.

API operations

Create and manage agents and runs; stream events, cancel work and download artifacts.

Request limits

API image input has a 15 MB limit; web attachments follow separate limits.

Retries

Use increasing delays for HTTP 429 and spread requests over time.

API versions

v1 is a public beta; v0 remains documented as a legacy interface.

Ownership and deployment

Keep your project files

Work stays in normal project source files and Git changes. Review, commit and move those files through your existing repository workflow.

Product open source status

Cursor is a licensed service and client; project source generated with it remains separate.

Export scope

Project source, Git changes and pull requests.

Privacy and organization controls

Control tool execution

MCP tools request approval by default. Enterprise administrators can restrict servers, tools and network destinations; distribution and permission approval are separate controls.

Privacy Mode

Privacy Mode is available in settings to free and paid users. Team members inherit the team setting, and model-provider protections apply when it is enabled.

Account deletion

Delete your account from the Settings dashboard. Support is available at hi@cursor.com for deletion assistance.

Model training

Inputs and suggestions are not used for training, with exceptions for security review, explicit feedback or training agreement.

Processing providers

Subprocessor information is published in the trust portal.

Security documents

Certification and attestation documents can be requested through the trust portal.

Export, migration and alternatives

Project and extension portability

Existing project files remain usable outside Cursor. Extension availability still needs checking: some Microsoft-only extensions use audited Anysphere alternatives.

Export exit options

Source files and Git history stay portable across development tools.

Sources and further reading

Information checked: 2026-10-02

Cursor official product

View source · Anysphere

Cursor pricing

View source · Anysphere

Pricing and plans

View source · Anysphere

Models & Pricing

View source · Anysphere

Cursor Download

View source · Anysphere

Cursor Security

View source · Anysphere

Cursor Privacy Policy

View source · Anysphere

Cursor Terms of Service

View source · Anysphere

Rollouts and Security Review

View source · Anysphere

Model Context Protocol (MCP)

View source · Anysphere, Inc.

Cloud Agents API

View source · Anysphere, Inc.

Cursor APIs Overview

View source · Anysphere, Inc.

Extensions

View source · Anysphere, Inc.

Take part in the discussion

Sign in to comment, save tools and rate them. You can edit or withdraw your own records. Saved tools are private. Comments are published after moderation and return to review after editing. Rating totals include only valid, approved ratings.

What can you do with it?

01

Implement a feature from project requirements.

02

Investigate an error and work through a fix.

03

Understand an unfamiliar repository.

04

Review a pull request before merging.

Related guides

Notion 3.6 brings external agents into the team workspaceProduct news · 4 min↗

Similar tools

Additional references

FindGoodAI current WordPress records and interaction implementation ↗

COMMUNITY

Tried it? Share your experience

Sign in to rate and comment
—Loading ratings
Your rating
One rating per person. Change it anytime.

User experiences

Loading experiences…

What did it help you do? What could be better?

Community and moderation

Saved tools are private. Approved comments are public; edits return to moderation. Comment counts include approved comments and replies. Each person has one rating. New ratings are automatically approved; withdrawn, unapproved or invalid ratings do not count. Share real experiences and avoid spam, private information or personal attacks.

For moderation appeals or data requests: support@findgoodai.com