• Custom roles vs static roles
• Technical overview
• How to implement a new ability for custom roles
  • Refactoring abilities
    • Finding existing abilities checks
    • Understanding logic for individual abilities
    • Abilities consolidation
  • Implement a new ability
• Custom abilities definition
  • Add a new custom ability definition
  • Schema
  • Privilege escalation consideration
  • Consuming seats

Custom Roles

Ultimate customers can create custom roles and define those roles by assigning specific abilities.

For example, a user could create an “Engineer” role with read code and admin merge requests abilities, but without abilities like admin issues.

In this context, the terms “permission” and “ability” are often used interchangeably.

• “Ability” is an action a user can do. These map to Declarative Policy abilities and live in Policy classes in ee/app/policies/*.
• “Permission” is how we refer to an ability in user-facing documentation. The documentation of permissions is manually generated so there is not necessarily a 1:1 mapping of the permissions listed in documentation and the abilities defined in Policy classes.

Custom roles vs static roles

In GitLab 15.9 and earlier, GitLab only had static roles as a permission system. In this system, there are a few predefined roles that are statically assigned to certain abilities. These static roles are not customizable by customers.

With custom roles, the customers can decide which abilities they want to assign to certain user groups. For example:

• In the static role system, reading of vulnerabilities is limited to a Developer role.
• In the custom role system, a customer can assign this ability to a new custom role based on any static role.

Like static roles, custom roles are inherited within a group hierarchy. If a user has custom role for a group, that user will also have a custom role for any projects or subgroups within the group.

Technical overview

• Individual custom roles are stored in the member_roles table (MemberRole model).
• A member_roles record is associated with top-level groups (not subgroups) via the namespace_id foreign key.
• A Group or project membership (members record) is associated with a custom role via the member_role_id foreign key.
• A Group or project membership can be associated with any custom role that is defined on the root-level group of the group or project.
• The member_roles table includes individual permissions and a base_access_level value.
• The base_access_level must be a valid access level. The base_access_level determines which abilities are included in the custom role. For example, if the base_access_level is 10, the custom role will include any abilities that a static Guest role would receive, plus any additional abilities that are enabled by the member_roles record by setting an attribute, such as read_code, to true.
• A custom role can enable additional abilities for a base_access_level but it cannot disable a permission. As a result, custom roles are “additive only”. The rationale for this choice is in this comment.
• Custom role abilities are supported at project level and group level.

How to implement a new ability for custom roles

Usually 2-3 merge requests should be created for a new ability. The rough guidance is following:

1. Pick a feature you want to add abilities to custom roles.
2. Refactor & consolidate abilities for the feature (1-2 merge requests depending on the feature complexity)
3. Implement a new ability (1 merge request)

Refactoring abilities

Finding existing abilities checks

Abilities are often checked in multiple locations for a single endpoint or web request. Therefore, it can be difficult to find the list of authorization checks that are run for a given endpoint.

To assist with this, you can locally set GITLAB_DEBUG_POLICIES=true.

This outputs information about which abilities are checked in the requests made in any specs that you run. The output also includes the line of code where the authorization check was made. Caller information is especially helpful in cases where there is metaprogramming used because those cases are difficult to find by grepping for ability name strings.

For example:

# example spec run

GITLAB_DEBUG_POLICIES=true bundle exec rspec spec/controllers/groups_controller_spec.rb:162

# permissions debug output when spec is run; if multiple policy checks are run they will all be in the debug output.

POLICY CHECK DEBUG -> policy: GlobalPolicy, ability: create_group, called_from: ["/gitlab/app/controllers/application_controller.rb:245:in `can?'", "/gitlab/app/controllers/groups_controller.rb:255:in `authorize_create_group!'"]

Use this setting to learn more about authorization checks while refactoring. You should not keep this setting enabled for any specs on the default branch.

Understanding logic for individual abilities

References to an ability may appear in a DeclarativePolicy class many times and depend on conditions and rules which reference other abilities. As a result, it can be challenging to know exactly which conditions apply to a particular ability.

DeclarativePolicy provides a ability_map for each policy class, which pulls all rules for an ability into an array.

For example:

> GroupPolicy.ability_map.map.select { |k,v| k == :read_group_member }
=> {:read_group_member=>[[:enable, #<Rule can?(:read_group)>], [:prevent, #<Rule ~can_read_group_member>]]}

> GroupPolicy.ability_map.map.select { |k,v| k == :read_group }
=> {:read_group=>
  [[:enable, #<Rule public_group>],
   [:enable, #<Rule logged_in_viewable>],
   [:enable, #<Rule guest>],
   [:enable, #<Rule admin>],
   [:enable, #<Rule has_projects>],
   [:enable, #<Rule read_package_registry_deploy_token>],
   [:enable, #<Rule write_package_registry_deploy_token>],
   [:prevent, #<Rule all?(~public_group, ~admin, user_banned_from_group)>],
   [:enable, #<Rule auditor>],
   [:prevent, #<Rule needs_new_sso_session>],
   [:prevent, #<Rule all?(ip_enforcement_prevents_access, ~owner, ~auditor)>]]}

DeclarativePolicy also provides a debug method that can be used to understand the logic tree for a specific object and actor. The output is similar to the list of rules from ability_map. But, DeclarativePolicy stops evaluating rules after you prevent an ability, so it is possible that not all conditions are called.

Example:

policy = GroupPolicy.new(User.last,  Group.last)
policy.debug(:read_group)

- [0] enable when public_group ((@custom_guest_user1 : Group/139))
- [0] enable when logged_in_viewable ((@custom_guest_user1 : Group/139))
- [0] enable when admin ((@custom_guest_user1 : Group/139))
- [0] enable when auditor ((@custom_guest_user1 : Group/139))
- [14] prevent when all?(~public_group, ~admin, user_banned_from_group) ((@custom_guest_user1 : Group/139))
- [14] prevent when needs_new_sso_session ((@custom_guest_user1 : Group/139))
- [16] enable when guest ((@custom_guest_user1 : Group/139))
- [16] enable when has_projects ((@custom_guest_user1 : Group/139))
- [16] enable when read_package_registry_deploy_token ((@custom_guest_user1 : Group/139))
- [16] enable when write_package_registry_deploy_token ((@custom_guest_user1 : Group/139))
  [21] prevent when all?(ip_enforcement_prevents_access, ~owner, ~auditor) ((@custom_guest_user1 : Group/139))

=> #<DeclarativePolicy::Runner::State:0x000000015c665050
 @called_conditions=
  #<Set: {
   "/dp/condition/GroupPolicy/public_group/Group:139",
   "/dp/condition/GroupPolicy/logged_in_viewable/User:83,Group:139",
   "/dp/condition/BasePolicy/admin/User:83",
   "/dp/condition/BasePolicy/auditor/User:83",
   "/dp/condition/GroupPolicy/user_banned_from_group/User:83,Group:139",
   "/dp/condition/GroupPolicy/needs_new_sso_session/User:83,Group:139",
   "/dp/condition/GroupPolicy/guest/User:83,Group:139",
   "/dp/condition/GroupPolicy/has_projects/User:83,Group:139",
   "/dp/condition/GroupPolicy/read_package_registry_deploy_token/User:83,Group:139",
   "/dp/condition/GroupPolicy/write_package_registry_deploy_token/User:83,Group:139"}>,
 @enabled=false,
 @prevented=true>

Abilities consolidation

Every feature added to custom roles should have minimal abilities. For most features, having read_* and admin_* should be enough. You should consolidate all:

• View-related abilities under read_*. For example, viewing a list or detail.
• Object updates under admin_*. For example, updating an object, adding assignees or closing it that object. Usually, a role that enables admin_ has to have also read_ abilities enabled. This is defined in requirement option in the ALL_CUSTOMIZABLE_PERMISSIONS hash on MemberRole model.

There might be features that require additional abilities but try to minimize those. You can always ask members of the Authentication and Authorization group for their opinion or help.

This is also where your work should begin. Take all the abilities for the feature you work on, and consolidate those abilities into read_, admin_, or additional abilities if necessary.

Many abilities in the GroupPolicy and ProjectPolicy classes have many redundant policies. There is an epic for consolidating these Policy classes. If you encounter similar permissions in these classes, consider refactoring so that they have the same name.

For example, you see in GroupPolicy that there is an ability called read_group_security_dashboard and in ProjectPolicy has an ability called read_project_security_dashboard. You’d like to make both customizable. Rather than adding a row to the member_roles table for each ability, consider renaming them to read_security_dashboard and adding read_security_dashboard to the member_roles table. This is more expected because it means that enabling read_security_dashboard on the parent group will enable the custom For example, GroupPolicy has an ability called read_group_security_dashboard and ProjectPolicy has an ability called read_project_security_dashboard. If you would like to make both customizable, rather than adding a row to the member_roles table for each ability, consider renaming them to read_security_dashboard and adding read_security_dashboard to the member_roles table. This convention means that enabling read_security_dashboard on the parent group will allow the custom role to access the group security dashboard and the project security dashboard for each project in that group. Enabling the same permission on a specific project will allow access to that projects’ security dashboard.

Implement a new ability

To add a new ability to a custom role:

• Generate YAML file by running ./ee/bin/custom-ability generator
• Add a new column to member_roles table, for example in this change in merge request 114734.

• Add the ability to the MemberRole model, ALL_CUSTOMIZABLE_PERMISSIONS hash, for example in this change in merge request 121534. There are following possible keys in the ALL_CUSTOMIZABLE_PERMISSIONS hash:

  • description - description of the ability.
  • minimal_level - minimal level a user has to have in order to be able to be assigned to the ability.
  • requirement - required ability for the ability defined in the hash, in case the requirement is false, the ability can not be true.
• Add the ability to the respective Policy for example in this change in merge request 114734.
• Update the specs.

Examples of merge requests adding new abilities to custom roles:

• Read code
• Read vulnerability
• Admin vulnerability - this is the newest MR implementing a new custom role ability. Some changes from the previous MRs are not necessary anymore (such as a change of the Preloader query or adding a method to User model).

You should make sure a new custom roles ability is under a feature flag.

Custom abilities definition

All new custom abilities must have a type definition stored in ee/config/custom_abilities that contains a single source of truth for every ability that is part of custom roles feature.

Add a new custom ability definition

To add a new custom ability:

1. Create the YAML definition. You can either:
   • Use the ee/bin/custom-ability CLI to create the YAML definition automatically.
   • Perform manual steps to create a new file in ee/config/custom_abilities/ with the filename matching the name of the ability name.
2. Add contents to the file that conform to the schema defined in ee/config/custom_abilities/types/type_schema.json.

Schema

Privilege escalation consideration

A base role typically has permissions that allow creation or management of artifacts corresponding to the base role when interacting with that artifact. For example, when a Developer creates an access token for a project, it is created with Developer access encoded into that credential. It is important to keep in mind that as new custom permissions are created, there might be a risk of elevated privileges when interacting with GitLab artifacts, and appropriate safeguards or base role checks should be added.

Consuming seats

If a new user with a role Guest is added to a member role that includes enablement of an ability that is not in the CUSTOMIZABLE_PERMISSIONS_EXEMPT_FROM_CONSUMING_SEAT array, a seat is consumed. We simply want to make sure we are charging Ultimate customers for guest users, who have “elevated” abilities. This only applies to billable users on SaaS (billable users that are counted towards namespace subscription). More details about this topic can be found in this issue.