Overview
Controls
Master capability toggles for features like web search, memory, and tools
togglescapabilitiesgovernance
Props
| Prop | Type | Default | Description |
|---|---|---|---|
controls | ControlCapability[] | required | Array of capability controls with id, label, optional description, enabled state, optional locked state, and optional status |
onToggleControl | (controlId: string, enabled: boolean) => void | required | Callback fired when a capability toggle changes |
title | string | "Controls" | Optional heading for the capability control panel |
variant | "list" | "cards" | "list" | Display style for the controls UI |
showStatus | boolean | true | Whether to display capability status badges/tags |
Docs
Controls provide a centralized set of master toggles for AI capabilities such as web search, memory, tool execution, or external actions. This pattern gives users explicit agency over what the AI can and cannot do, making capability boundaries visible and adjustable in real time.
Variants
- Capability list — A vertical list of switches with labels and descriptions for each AI capability.
- Card grid — A denser, card-like presentation suitable for dashboards with many controls.
- Policy-locked controls — Some toggles are visible but locked to communicate organization or admin policy constraints.
Use Cases
- Turning web browsing on only when fresh data is required
- Disabling memory for sensitive or one-off sessions
- Restricting code execution to trusted workflows
- Allowing users to opt out of external tool actions
- Enterprise policy surfaces where some controls are admin-managed
Best Practices
- Group controls by risk level so users can reason about impact quickly.
- Use concise, plain-language descriptions of what each capability does.
- Show locked states clearly and explain why a control cannot be changed.
- Apply control changes immediately and provide visible confirmation.
- Pair critical capability toggles with audit trails in high-stakes products.