Cloudflare Access determines who can reach your application by applying the Access policies you configure.
An Access policy consists of an Action as well as rules which determine the scope of the action. To build a rule, you need to choose a Rule type, Selector, and a Value for the selector.
Actions let you grant or deny permission to a certain user or user group. You can set only one action per policy.
The Allow action allows users that meet certain criteria to reach an application behind Access.
The following example lets any user with an
@example.com email address, as validated against an IdP, reach the application:
|Allow||Include||Emails Ending In:|
You can add a Require rule in the same policy action to enforce additional checks. Finally, if the policy contains an Exclude rule, users meeting that definition are prevented from reaching the application.
For example, this second configuration lets any user from Portugal with a
@team.com email address, as validated against an IdP, reach the application, except for
|Require||Emails Ending In|
The Block action prevents users from reaching an application behind Access.
For example, this configuration blocks every request to the application, except for requests from
The Bypass action disables any Access enforcement for traffic that meets the defined rule criteria. Bypass is typically used to enable applications that require specific endpoints to be public. For example, some applications have an endpoint under the
/admin route that must be publicly routable. In this situation, you could create an Access application for the domain
test.example.com/admin/<your-url> and add the following Bypass policy:
As part of implementing a Zero Trust security model, we do not recommend using Bypass to grant direct permanent access to your internal applications. To enable seamless and secure access for on-network employees, use Cloudflare Tunnel to and have users connect through WARP.
Service Auth rules enforce authentication flows that do not require an identity provider IdP login, such as service tokens and mutual TLS.
|Service Auth||Include||Valid certificate|
Rules work like logical operators. They help you define which categories of users your policy will affect.
|The Include rule is similar to an OR logical operator. In case more than one Include rule is specified, users need to meet only one of the criteria.||The Exclude rule works like a NOT logical operator. A user meeting any Exclusion criteria will not be allowed access to the application.||The Require rule works like an AND logical operator. A user must meet all specified Require rules to be allowed access.|
All Access policies must contain an Include rule. This is what defines the initial pool of eligible users who can access an application. You can then add Exclude and Require rules to enforce specific policies for those users.
Requiring multiple conditions
When setting up a Require rule for an Access policy, keep in mind that any values you add to the rule will be concatenated by an AND operator. For example, let’s say you want to grant access to an application to both the full-time employees and the contractors, and only the ones based in specific countries — say Portugal and the United States. If you set up a rule with the following configuration:
|Allow||Require||Emails ending in|
the policy will only grant access to people reaching the application from both the United States AND Portugal, and who have both an email ending in
@cloudflare.com AND in
@contractors.com. Therefore, nobody will have access to the application.
Next, you can create a policy for your application that requires the group, and that also includes users with emails ending in either
|Include||Emails ending in|
When you add a rule to your policy, you will be asked to specify the criteria you want users to meet.
These criteria are available for all Access application types, including , , and applications. Identity-based attributes are only checked when a user authenticates, whereas other attributes are polled continuously for changes during the session.
|Selector||Description||Checked at login||Checked continuously|
|Emails ending in||✅||❌|
|External Evaluation||Allows or denies access based on in an external API.||✅||❌|
|Country||Uses the IP address to determine country.||✅||✅|
|Everyone||Allows, denies, or bypasses access to everyone.||✅||❌|
|Common Name||The request will need to present a valid certificate with an expected common name.||✅||✅|
|Valid Certificate||The request will need to present any valid client certificate.||✅||✅|
|Service Token||The request will need to present the correct service token headers configured for the specific application.||✅||✅|
|Any Access Service Token||The request will need to present the headers for any created for this account.||✅||✅|
|Login Methods||Checks the identity provider used at the time of login.||✅||❌|
|Authentication Method||Checks the method used by the user, if supported by the identity provider.||✅||❌|
|Identity provider group||Checks the user groups you configured with your identity provider (IdP). This selector only displays if you use AzureAD, GitHub, Google, or Okta as your IdP.||✅||❌|
|SAML Group||Checks a SAML attribute name / value pair. This selector only displays if you use a identity provider.||✅||❌|
|OIDC Claim||Checks an OIDC claim name / value pair. This selector only displays if you use a identity provider.||✅||❌|
|Device posture||Checks from the WARP client or a third-party service provider.||✅||✅|
|Warp||Checks that the device is connected to WARP, including the consumer version.||✅||✅|
|Gateway||Checks that the device is connected to your Zero Trust instance through the .||✅||✅|
Order of execution
Policies are evaluated based on their action type and ordering. Bypass and Service Auth policies are evaluated first, from top to bottom as shown in the UI. Then, Block and Allow policies are evaluated based on their order.
For example, if you have a list of policies arranged as follows:
- Allow A
- Block B
- Service Auth C
- Bypass D
- Allow E
The policies will execute in this order: Service Auth C > Bypass D > Allow A > Block B > Allow E. Once a user matches an Allow or Block policy, evaluation stops and no subsequent policies can override the decision.