How to block Cursor in your organisation

Cursor is an AI code editor with an agent that edits files, runs terminal commands and connects to MCP servers. Block it on devices where it is not approved, or allow only your company's Cursor team, and know which "blocked" message comes from whom.

Next step

AI Agent Blocklist: Cursor and every other AI code editor and coding agent, as app IDs, process names, package names and domains

Ready for Intune, Jamf, Defender, your firewall and your DNS filter, with a risk grade on each entry and daily updates.

From $99 a month for one organisation See plans Ask a question

Why IT teams block or restrict Cursor

Cursor looks like a familiar code editor, so it spreads quickly. Underneath, it is an agent platform with access to the code, the terminal and outside services.

Code leaves the device

The AI features send code context to Cursor's servers and model providers. Privacy settings change what is kept, not what is sent.

The agent runs commands

Agent mode can run terminal commands. Auto-run settings decide whether a person approves each one.

MCP servers

Developers can add MCP servers that bring new tools and data sources, each one code you did not review.

Personal accounts

Free, trial and personal Pro accounts sit outside your contract, your SSO and your data terms.

Ignore files are not walls

Cursor's documentation says .cursorignore is a convenience, not a security boundary. Terminal commands and MCP tools can still read ignored files.

Injected instructions

Text the agent reads in code, issues or web pages can carry instructions. Treat MCP configuration files as sensitive and review every change to them, as you would a build script.

The wider picture, with the risk patterns seen across coding agents, is in AI coding agent security risks. For the MCP side, see MCP security risks.

Signs Cursor is already on your network

Most IT teams find Cursor after it is already in daily use. Collect these signals first, so the block does not surprise a team that depends on it.

On devices

  • The Cursor app in macOS and Windows software inventory
  • The cursor-agent process started from terminals
  • A .cursor folder in home directories and projects
  • .cursor/mcp.json files that define MCP servers

In repositories

  • Rules files for the Cursor agent committed to repos
  • .cursorignore files next to code
  • Commits created by agent sessions
  • New MCP configuration in pull requests

On the network

  • DNS queries for cursor.com and cursor.sh
  • Regular traffic from developer subnets to AI model providers
  • Sign-ins with personal email addresses
  • Help desk tickets about blocked AI editors

"Cursor blocked me": who blocked it and why

Users search for these messages when Cursor stops working. Some are your policy, some are Cursor's own checks, and some are endpoint protection. The help desk needs to know which is which.

What the user sees or searchesWho blocked itWhat is going onWhat to do
Cursor blocked by team admin, logged out right after sign-inYour ITThe AllowedTeamId device policy only accepts listed team IDs. Other accounts are logged out and further attempts are blocked.Sign in with the company team account
Access denied on the company network, tenant restriction errorYour ITThe proxy adds the x-cursor-allowed-tenants header, and Cursor refuses accounts outside the listed teams or organisations.Use the company account; off the network the same account is not limited
Your request has been blocked as our system has detected suspicious activityCursorCursor's anti-abuse system, which its staff say mostly targets free and trial accounts.Use a company seat; contact Cursor support if it persists
Cursor blocked by Windows DefenderEndpoint protectionA Defender or app control rule blocks the app on purpose, or a reputation check flagged a download.Check your Defender and Intune policies first
Cursor blocked access, cannot reach the serverYour ITThe firewall, DNS filter or gateway blocks Cursor's domains.Intended if Cursor is not approved
An extension will not install or run in CursorLicence or protectionSome extensions are licensed only for specific editors, or a Defender rule blocks the download, as above.Check the extension's licence terms and your Defender rules
Never help users get around Cursor's anti-abuse checks with machine ID tools found online. They change device identifiers and break the vendor's terms. The fix inside a company is a company seat.

Cursor identifiers and domains

Cursor ships as a desktop app and a command-line agent. A few of the identifiers in our files are below; the files also cover helpers, companion tools and every other AI editor we track.

Desktop app

  • macOS bundle IDcom.todesktop.230313mzl4w4u92
  • WinGet package IDAnysphere.Cursor
  • macOS processCursor
  • Homebrew caskcursor

CLI and network

  • Cursor CLI processcursor-agent
  • Homebrew cask of the CLIcursor-cli
  • Main domaincursor.com
  • Service domaincursor.sh

The macOS bundle ID looks unusual because the app is packaged with a third-party build tool. Use it exactly as shown; a guessed ID such as one based on the product name will not match.

Device rules for Cursor on Mac and Windows

Network blocks stop the AI features; device rules stop the app. Use the tool you already manage endpoints with, and test on one machine before you widen the scope.

PlatformToolWhat to matchNotes
macOSJamf Pro, Kandji, Mosyle or another MDMBundle ID and process name in a restricted software or app block ruleAlso block the Homebrew casks if developers use Homebrew
WindowsIntune with App Control for Business or AppLocker rulesThe publisher and executable of the Cursor appAdd the WinGet package ID to your blocked packages if you manage WinGet
WindowsMicrosoft Defender for EndpointFile or certificate indicators for the installer and appGives a clear block message in Defender alerts
LinuxYour configuration managementThe AppImage or package and the cursor-agent binaryDevelopers often install in their home folder, so check user paths
Tell the help desk before the rule lands. Users who see a Defender or app control message will search for "Cursor blocked by Windows Defender", and a short note on your intranet answers it.

Two ways to handle Cursor

Decide first whether Cursor is approved for anyone. The controls for "nobody" and "only our team" are different.

Block Cursor for everyone

  1. Device rules: block the app by bundle ID on Macs and by package ID and executable on Windows, with Jamf, Intune or Defender.
  2. CLI: block the cursor-agent process and the Homebrew casks.
  3. Network: add cursor.com and cursor.sh to your firewall or DNS block list. Sign-in and AI features stop.
  4. Other AI editors: block them in the same change, or developers move to the next one.
  5. Message: tell developers which editor and AI tools are approved.

Allow only your company team

  1. Plan: the device and identity controls below are Enterprise features; check your plan.
  2. SSO: require SAML sign-in for all team members and block password sign-in.
  3. Device policy: deploy AllowedTeamId with your team IDs through MDM.
  4. Network: add the tenant header on the proxy for requests to Cursor's domains.
  5. Admin settings: set privacy, auto-run, MCP and model rules in the team dashboard.

What each kind of Cursor account gives IT

The account type decides how much control you have. This is why "allow Cursor" without a company team usually means "allow personal accounts".

AccountWho controls itControls IT gets, per Cursor's dashboard documentation
Free, trial or personal ProThe developerNone. Your data terms and SSO do not apply.
TeamsYour team adminsCentral billing, team privacy settings, usage-based pricing settings
EnterpriseYour team admins and ITAdds MDM policies such as AllowedTeamId, Auto Run Configuration, MCP Configuration, Repository Blocklist, model controls and SCIM

Restrict Cursor to your team: the three locks

Each lock covers a gap the others leave. Device policy follows the laptop, the network header covers anything on your network, and SSO decides who belongs to the team in the first place.

1. Device policy

The AllowedTeamId policy, delivered by MDM, overrides the cursorAuth.allowedTeamId setting on the device. Its value is a comma-separated list of team IDs.

AllowedTeamId = "1234,5678"

On Windows, the ADMX and ADML files ship in the Cursor install folder under AppData\Local\Programs\cursor\policies for Group Policy.

2. Network header

Your proxy adds a header on requests to cursor.com and the API host under cursor.sh. Accounts outside the list are refused with a tenant restriction violation.

x-cursor-allowed-tenants: v1.orgs=1001;teams=70

The restriction is not stored on the token, so it applies only on your network.

3. Identity

Enforce SAML SSO for the team and use SCIM provisioning, available on Enterprise with SSO, so leavers lose access when they leave your directory.

SSO required, password sign-in off

Check the current admin dashboard for exact labels.

Two more MDM policies are worth setting while you are there: AllowedExtensions, which controls which extensions may be installed, and WorkspaceTrustEnabled, which keeps workspace trust on. Both are listed in Cursor's enterprise identity documentation.

Team dashboard settings to review

Once only your team can sign in, the dashboard decides what the agent may do. These are the settings with the most security weight. Names can change, so check the current admin dashboard.

  • Privacy Settings Teams and EnterpriseEnforce privacy mode for the whole team instead of leaving it to each user.
  • Auto Run Configuration EnterpriseDecide whether the agent may run terminal commands without asking.
  • MCP Configuration EnterpriseAllow MCP servers by command or URL, limit their tools, and set per-server network modes such as Allowlist or Deny all.
  • Repository Blocklist EnterpriseStop the editor and CLI from indexing or working with listed repositories.
  • Models EnterpriseChoose which models members may use, and set defaults.
  • Personal API Keys (BYOK)Turn off members' own provider keys so all use runs under the team's terms.
MCP rules set in the dashboard override the user's local permissions file and editor settings. For a method to decide which servers make the cut, read building an MCP allowlist.

Keeping secrets away from Cursor's agent

Whether you allow Cursor or not, secrets on developer machines are the prize. Treat the ignore file as a hint and put real controls around it.

Remove

  • No production secrets in project folders
  • Short-lived tokens loaded at run time
  • Secret scanning on every commit

Require approval

  • Terminal commands need a person to approve
  • Network calls from the agent are reviewed
  • New MCP servers need IT approval

Watch

  • Agent activity in endpoint logs
  • Outbound traffic to unknown AI domains
  • Repository changes made by agents
Side-by-side settings for Cursor, Claude Code and Codex are in stop AI coding agents from reading .env files and secrets.

What to tell developers

A clear message prevents most workarounds. Keep it short and send it before the block applies.

  • What changesCursor is blocked on company devices, or allowed only with the company team account, from a given date.
  • WhyCode and credentials must stay under company terms, and agent actions need company rules.
  • What to use insteadThe approved editor and AI assistant, with a link to how to get access.
  • How to ask for an exceptionWho decides, what information they need, and how long a decision takes. Link your AI agent use policy.

How the AI Agent Blocklist covers Cursor and other AI editors

Cursor is one of many AI code editors and coding agents. If you block one, the next appears on a developer's laptop within a week. Our files cover the category with one import.

Where you enforceWhat our files give you
Jamf, Kandji, MosylemacOS bundle IDs and process names of AI editors and agent apps, see Jamf on Macs
Intune and DefenderWinGet package IDs and Windows process names, see block AI with Intune and Microsoft Defender
Firewall and DNS filterDomains of AI editors, coding agents and MCP services, in plain, hosts, PAC and firewall list formats
Package proxiesnpm and PyPI names of command-line coding agents

Block the category, not just one editor

The AI Agent Blocklist covers AI code editors, coding agents, MCP servers, AI browsers and agent extensions in one subscription.

From $99 a month for one organisation. Download access is set up within one business day of your order.

See plans Ask a question
  • macOS bundle IDs and Windows package IDs for MDM
  • Process names for endpoint rules
  • Package names of command-line coding agents
  • 2,600+ AI agent domains in firewall and DNS formats
  • Risk grade on every entry, updated every day

Ask us or order a plan

Tell us what you need to block and where your rules live. We reply within one business day.

  • Files for firewalls, DNS filters, secure web gateways, MDM and browser policies
  • Updated every day as new AI browsers, agents and extensions appear
  • Plans for one organisation, for MSPs and for product makers, or a hosted filter we run for you

We use your details only to answer you. No newsletters.

Questions about blocking Cursor

How do I block Cursor on company computers?
Block the app on devices by its macOS bundle ID, WinGet package ID and process names, block the Cursor CLI, and block the Cursor domains on your firewall or DNS filter. If some teams may use Cursor, restrict sign-in to your company team instead of blocking everyone.
What does Cursor blocked by team admin mean?
Your organisation restricted which Cursor teams may sign in on that device or network, usually with the AllowedTeamId device policy or a tenant restriction header on the company proxy. Sign in with your company account, or ask IT to add the team.
Why does Cursor say my request has been blocked due to suspicious activity?
That message comes from Cursor's own anti-abuse checks, not from your IT team. Cursor staff have said these checks mostly affect free and trial accounts. Company users should use the company's Cursor team account and contact Cursor support if the message persists.
Can I allow Cursor only for our company team?
Yes. On Cursor Enterprise, deploy the AllowedTeamId policy with your team IDs through MDM so other accounts are logged out, and have your proxy add the x-cursor-allowed-tenants header so other accounts are refused on the network. Enforce SSO for the team as well.
Why is Cursor blocked by Windows Defender?
Either a Microsoft Defender rule in your organisation blocks the app on purpose, or SmartScreen or another protection flagged a download it did not recognise. Check your Defender and Intune app control policies first; if the block is not intended, verify the installer came from the vendor.
Is .cursorignore enough to keep secrets away from Cursor's agent?
No. Cursor's own documentation describes the ignore file as a convenience, not a security boundary, and terminal commands and MCP tools can still read ignored files. Keep secrets out of the workspace and require approval before terminal commands.
Does blocking Cursor's domains stop the editor?
It stops sign-in and the AI features, which rely on Cursor's servers. The editor may still open for plain editing. To stop the app itself, add a device rule for the app and its processes.

Cursor is a trademark of Anysphere, Inc. Claude Code is a trademark of Anthropic, PBC. Codex is a trademark of OpenAI. GitHub Copilot is a trademark of GitHub, Inc.

Microsoft Defender, Intune, SmartScreen and Windows are trademarks of Microsoft Corporation. Jamf, Kandji and Mosyle are trademarks of their respective owners.

Apple, Mac and macOS are trademarks of Apple Inc. WinGet and AppLocker are trademarks of Microsoft Corporation. Linux is a registered trademark of Linus Torvalds.

npm is a trademark of npm, Inc. PyPI is a trademark of the Python Software Foundation. Homebrew is a trademark of its owner.

Names are used only to identify the products discussed. AI Agent Blocklist is not affiliated with, sponsored or endorsed by these companies.