Build Agentic workflows, RAG pipelines, with rich AI model and tool support on one collaborative workspace. Deploy on cloud, VPC, or self-hosted, so teams move from prototype to production without rebuilding the stack.
Python agentagentic-aiagentic-frameworkagentic-workflowaiautomationclaudedeepseekgenaigptllmlow-codemcpnextjsno-codeopenaipythonskillsworkflow
import type { HumanInputRequiredResponse } from '@/types/workflow'
import { produce } from 'immer'
import { useCallback } from 'react'
import { useStoreApi } from 'reactflow'
import { updatePendingHumanInputForm } from '@/app/components/base/chat/chat/answer/human-input-content/form-state'
import { useWorkflowStore } from '@/app/components/workflow/store'
import { NodeRunningStatus } from '@/app/components/workflow/types'
export const useWorkflowNodeHumanInputRequired = () => {
const store = useStoreApi()
const workflowStore = useWorkflowStore()
// Notice: Human input required !== Workflow Paused
const handleWorkflowNodeHumanInputRequired = useCallback(
(params: HumanInputRequiredResponse) => {
const { data } = params
const { workflowRunningData, setWorkflowRunningData } = workflowStore.getState()
const newWorkflowRunningData = produce(workflowRunningData!, (draft) => {
updatePendingHumanInputForm(draft, data)
const currentIndex = draft.tracing!.findIndex((item) => item.node_id === data.node_id)
if (currentIndex > -1) {
draft.tracing![currentIndex] = {
...draft.tracing![currentIndex]!,
status: NodeRunningStatus.Paused,
}
}
})
setWorkflowRunningData(newWorkflowRunningData)
const { getNodes, setNodes } = store.getState()
const nodes = getNodes()
const currentNodeIndex = nodes.findIndex((node) => node.id === data.node_id)
const newNodes = produce(nodes, (draft) => {
draft[currentNodeIndex]!.data._runningStatus = NodeRunningStatus.Paused
})
setNodes(newNodes)
},
[store, workflowStore],
)
return {
handleWorkflowNodeHumanInputRequired,
}
}