Skip to main content
An environment is the surface an agent perceives and acts upon. H focuses on visual computer-use: the agent looks at the screen, reasons, and acts, step after step. An agent gets its environments from its environments list, where each entry is either a catalog name or an inline environment object. At least one is required, with at most one per kind.
Beta: only the Browser environment is enabled today. More kinds, including desktop environments for Mac, Windows, and Linux, are in What’s next.

​
Toolbox

Attaching an environment equips the agent with a toolbox: a fixed action surface (the toolset it calls to operate the environment) plus built-in skills injected into the system prompt so it knows how to use them. The Browser, for example, contributes its web actions (navigate, click, type, scroll, read) and the matching operating guidance. The action surface is fixed per environment; you cannot add or remove individual actions. To shape how the agent uses them, set the agent’s instructions or attach your own skills, which load on demand on top of the built-in ones.

​
Built-in environments

H ships ready-to-use browsers in the catalog. Browse them below, or list them anytime with GET /api/v2/environments:

​
Endpoints

MethodPathDescription
POST/api/v2/environmentsCreate an environment
GET/api/v2/environmentsList environments
GET/api/v2/environments/{id}Retrieve an environment
PUT/api/v2/environments/{id}Update an environment
DELETE/api/v2/environments/{id}Delete an environment