Boundary's API is a JSON-based HTTP API that adheres to a set of standards that are rigidly followed. At its core, it is a standards-compliant JSON API for both input and output.
Before reading this page, it is useful to understand Boundary's domain model to be aware of the terminology used here.
Boundary's API is also described via OpenAPI v2; the version corresponding to any tag of Boundary's source code can be found in Boundary's GitHub repository.
Boundary's current API version is 1; all API paths begin with
2XX: Boundary returns a code between
299on success. Generally this is
200, but implementations should be prepared to accept any
2XXstatus code as indicating success. If a call returns a
2XXcode that is not
200, it will follow well-understood semantics for those status codes. (Starting with Boundary 0.2.1,
400: Boundary returns
400when a command cannot be completed due to invalid user input, except for a properly-formatted identifier that does not map to an existing resource, which will return a
404as discussed below.
401: Boundary returns
401if no authentication token is provided or if the provided token is invalid. A valid token that simply does not have permission for a resource will return a
403instead. A token that is invalid or missing, but where the anonymous user (
u_anon) is able to successfully perform the action, will not return a
401but instead will return the result of the action.
403: Boundary returns
403if a provided token was valid but does not have the grants required to perform the requested action.
404: Boundary returns
404if a resource cannot be found. Note that this happens prior to authentication/authorization checking in nearly all cases as the resource information (such as its scope, available actions, etc.) is a required part of that check. As a result, an action against a resource that does not exist will return a
404instead of a
403. While this could be considered an information leak, since IDs are randomly generated and this only discloses whether an ID is valid, it's tolerable as it allows for far simpler and more robust client implementation.
405: Boundary returns a
405to indicate that the method (HTTP verb or custom action) is not implemented for the given resource.
500: Boundary returns
500if an error occurred that is not (directly) tied to invalid user input. If a
500is generated, information about the error will be logged to Boundary's server log but is not generally provided to the client.
Boundary follows a predictable path layout. There are two fundamental types of URL paths, each supporting a different set of operations.
Collections of resources are top level paths with plural English names for the resource, e.g.
/hosts. Collections support the following operations:
- Creation of new resources within that collection
- Listing of resources within that collection
All collection operations require supplying the enclosing resource. Depending on the collection type, this will be one of the following:
- A scope, indicating the scope in which an operation should take place. For instance, a POST to
/roleswould need to indicate whether the role should be created within the
globalscope or an org-level scope like
- A parent resource of the appropriate type. For instance, hosts and host sets are child resources of host catalogs. When creating a new host set within a host catalog, a POST to
/host-setswould need to indicate the host catalog ID with which the host-set should be associated.
Resources themselves are defined by ID specifiers within a collection path, e.g.
/roles/r_1234567890. Resources support the following operations:
- Reading a resource's properties
- Updating a resource's properties
- Deleting a resource
- Custom methods specific to a resource type
Depending on the resource type, various parameters may be available. Some are common across all resource types (e.g.
description); others are available only for specific types. Further, some concrete-types of abstract resources include an opaque
attributes JSON object with type-specific values.
For instance, an auth method is an abstract type; a
password auth method is a concrete implementation of that type. When creating such an auth method, a
type parameter will indicate that it should be the
password type, while values specific to the
password type auth method, such as minimum password length, will be contained within an
The following method conventions are used within Boundary's API:
GET is used for reading a resource or listing resources in a collection. The behavior depends on whether the
GET is issued against a collection (
/roles) or a singular resource (
/roles/r_1234567890). In the former case it lists resources within the collection; in the latter it performs a read on that particular resource.
POST is used for creating a resource or performing custom actions against a resource. When creating a resource,
POST is used against a collection (
/roles). When performing a custom action,
POST is used against a particular resource (
PATCH is used to update a resource's parameters. The following are behaviors to be aware of when using
- In nearly all cases, a
versionparameter is required. This is used for check-and-set, to ensure that the update operation is being performed against a known resource. The version parameter is returned from a
GEToperation on the resource so the current version, along with the resource's other current values, can be looked up at any time.
- Passing a JSON
nullfor a parameter has the effect of reverting that parameter to its default. For some parameters (e.g.
name) this will simply clear the value (as the default
namefor a resource is empty); for other parameters this will revert to the current defaults within Boundary.
- All parameters specified as part of a
PATCHoperation will be considered to be parameters that should be updated.
DELETE is used for deleting a specific resource, and is only used against a particular resource path.