HyperView docs
HyperView · Guides
CLI and agents
Drive a running HyperView workspace from the command line, so people and coding agents work on the same state.
Everything you see in a HyperView workspace — datasets, layouts, selections, panels, jobs — is state owned by the server, not by the browser tab. The hyperview CLI reads and changes that same state. A coding agent that runs CLI commands is therefore working in the workspace you are looking at, and you see its changes as they happen.
Workspaces
A workspace binds a dataset to a UI arrangement and its state.
hyperview workspace create imagenette-demo --dataset imagenette --activate
hyperview serve --workspace imagenette-demo --dataset imagenette
hyperview workspace list
Control the UI
hyperview ui layout set --workspace imagenette-demo --layout-key <layout-key>
hyperview ui selection set --workspace imagenette-demo --ids sample-1,sample-8
hyperview ui panel definitions
Run tools
Tools are Python functions contributed by extensions. They take JSON parameters and return JSON, so they work the same from a panel, a script or an agent:
hyperview extension add .hyperview/extensions/selection-profile --workspace imagenette-demo
hyperview tools list
hyperview tools run selection_profile.summarize --workspace imagenette-demo
Command groups
| Group | Purpose |
|---|---|
serve, status | Run a workspace server; check what is running |
dataset | Create, list and inspect datasets |
embeddings, layouts, jobs | Compute spaces and layouts; follow jobs |
provider | Register custom embedding providers |
workspace, ui, panel | Manage workspaces and what they show |
extension, tools | Install extensions; run their tools |
export, publish | Package and host a workspace as a Space |
skill | Install the HyperView agent skill |
commands | List and run the runtime's control commands |
Add --json to most commands for machine-readable output. hyperview <group> --help lists the options.
Set up your agent
hyperview skill install
The skill gives coding agents the command reference and the conventions above. Start a server, then ask your agent to, for example, "find the outliers in the hyperbolic layout and select them".