@sqlrooms/ai-core / ChatSuggestionsState
Interface: ChatSuggestionsState
Normalized prompt-suggestions state and actions, identical in shape across both chat runtime modes.
Read via usePromptSuggestions. This is the behavior layer: no DOM, no styling. send and isReadyToSend reuse the composer's own send action and readiness signals, so a suggestion and the send control never disagree.
Properties
mode
mode:
ChatComposerMode
Which runtime this state was sourced from.
visible
visible:
boolean
Whether suggestions are currently visible.
setVisible
setVisible: (
visible) =>void
Sets visibility directly.
Parameters
| Parameter | Type |
|---|---|
visible | boolean |
Returns
void
toggle
toggle: () =>
void
Flips visibility.
Returns
void
items
items: readonly
string[]
Runtime-supplied suggestion strings, possibly empty. Session mode has no concept of runtime-supplied suggestions and always reports an empty array; local-agent mode reports the root's initialSuggestions.
isSessionEmpty
isSessionEmpty:
boolean
True when the chat has no messages and no in-progress prompt, in both modes — including a draft typed before any session exists. Branch on this instead of a when-style prop.
fill
fill: (
text) =>void
Writes text into the prompt without sending it.
Parameters
| Parameter | Type |
|---|---|
text | string |
Returns
void
send
send: (
text) =>void
Sends text immediately, subject to isReadyToSend. Routed through the composer's send, so pre-send vetoes apply.
Parameters
| Parameter | Type |
|---|---|
text | string |
Returns
void
isReadyToSend
isReadyToSend:
boolean
True when an item's own text could be sent: a model is resolvable (session mode) and nothing is running or summarizing. Built from the same signals as useChatComposer's canSend, minus its non-empty-prompt requirement, which is the wrong question for an item supplying its own text.
Built on sendBlocked, not needsApiKey: the latter is also true for apps needing no browser key (remote chatEndPoint, no ai-settings slice).
