Menu
Language
Theme
Stream interface for DeepAgent instances created with createDeepAgent.
Extends UseAgentStream with subagent streaming capabilities. Subagent streams are automatically typed based on the agent's subagent configuration, enabling type-safe access to subagent state and messages.
Use this interface when streaming from an agent created with createDeepAgent
that orchestrates multiple specialized subagents.
This interface is subject to change.
interface UseDeepAgentStreamUseAgentStream<StateType, ToolCall, Bag>This interface adds subagent streaming on top of UseAgentStream:
subagents - Map of all subagent streams by tool call IDactiveSubagents - Array of currently running subagentsgetSubagent(id) - Get a specific subagent by tool call IDgetSubagentsByType(type) - Get all subagents of a specific type with typed stategetSubagentsByMessage(messageId) - Get all subagents triggered by a specific AI messageIt also enables the filterSubagentMessages option to exclude subagent
messages from the main messages array.
import { createDeepAgent } from "deepagents";
import { useStream } from "@langchain/langgraph-sdk/react";
// Define subagents with typed middleware
const agent = createDeepAgent({
subagents: [
{
name: "researcher",
description: "Research specialist",
middleware: [ResearchMiddleware],
},
{
name: "writer",
description: "Content writer",
middleware: [WriterMiddleware],
},
] as const, // Important: use 'as const' for type inference
});
// In React component:
function Chat() {
const stream = useStream<typeof agent>({
assistantId: "deep-agent",
apiUrl: "http://localhost:2024",
filterSubagentMessages: true, // Only show main agent messages
});
// Subagent streams are typed!
const researchers = stream.getSubagentsByType("researcher");
researchers.forEach(subagent => {
// subagent.values.messages is typed as Message<ToolCall>[]
// subagent.status is "pending" | "running" | "complete" | "error"
console.log("Researcher status:", subagent.status);
});
// Track all active subagents
stream.activeSubagents.forEach(subagent => {
console.log(`${subagent.toolCall.args.subagent_type} is running...`);
});
}Currently active subagents (where status === "running").
Use this to track and display subagents that are actively executing. Completed or errored subagents are not included.
// Show loading indicators for active subagents
stream.activeSubagents.map(subagent => (
<SubagentCard
key={subagent.id}
type={subagent.toolCall.args.subagent_type}
isLoading={true}
/>
));The ID of the assistant to use.
The current branch of the thread. Used for navigating between different conversation branches.
LangGraph SDK client used to send requests and receive responses.
Last seen error from the stream, if any.
Reset to undefined when a new stream starts.
Tree of all branches for the thread. This API is experimental and subject to change.
Get subagent stream by tool call ID.
Use this when you have a specific tool call ID and need to access its corresponding subagent stream.
// In a tool call component
const subagent = stream.getSubagent(toolCall.id);
if (subagent) {
return <SubagentProgress subagent={subagent} />;
}Get all subagents triggered by a specific AI message.
Useful for rendering subagent activities grouped by conversation turn. Each AI message that contains subagent tool calls will have its triggered subagents returned by this method.
// Render subagents inline after the AI message that triggered them
{stream.messages.map((msg) => (
<div key={msg.id}>
<MessageBubble message={msg} />
{msg.type === "ai" && "tool_calls" in msg && (
<SubagentPipeline
subagents={stream.getSubagentsByMessage(msg.id)}
/>
)}
</div>
))}Get all subagents of a specific type.
Returns streams with properly inferred state types based on subagent name. When called with a literal string that matches a subagent name, TypeScript will infer the correct state type for that subagent.
// Get all researcher subagents with typed state
const researchers = stream.getSubagentsByType("researcher");
researchers.forEach(researcher => {
// researcher.values is typed based on ResearchMiddleware
console.log("Research messages:", researcher.values.messages.length);
console.log("Status:", researcher.status);
});
// Get all writer subagents
const writers = stream.getSubagentsByType("writer");
// writers have different state type based on WriterMiddlewareGet tool calls for a specific AI message.
Use this to find which tool calls were initiated by a particular assistant message, useful for rendering tool calls inline with messages.
messages.map(message => {
if (message.type === "ai") {
const calls = stream.getToolCalls(message);
return (
<>
<MessageBubble message={message} />
{calls.map(tc => <ToolCallCard key={tc.call.id} {...tc} />)}
</>
);
}
return <MessageBubble message={message} />;
});Flattened history of thread states of a thread. Contains all states in the current branch's history.
Current interrupt, if the stream is interrupted.
Convenience alias for interrupts[0].
For workflows with multiple concurrent interrupts, use interrupts instead.
All current interrupts from the stream. When using Send() fan-out with per-task interrupt() calls, multiple interrupts may be pending simultaneously.
Whether the stream is currently running.
true while streaming, false when idle or completed.
Whether the thread is currently being loaded.
true during initial thread data fetch.
Set the branch of the thread.
Stops the currently running stream.
All currently active and completed subagent streams.
Keyed by tool call ID for easy lookup. Includes subagents in all states: pending, running, complete, and error.
// Iterate over all subagents
stream.subagents.forEach((subagent, toolCallId) => {
console.log(`Subagent ${toolCallId}: ${subagent.status}`);
});
// Get a specific subagent
const specific = stream.subagents.get("call_abc123");Tool calls paired with their results.
Each entry contains the tool call request and its corresponding result. Useful for rendering tool invocations and their outputs together.
stream.toolCalls.map(({ call, result }) => (
<ToolCallCard
name={call.name}
args={call.args}
result={result}
/>
));The current state values of the stream. Updated as streaming events are received.
The ID of the assistant to use.
The current branch of the thread.
LangGraph SDK client used to send requests and receive responses.
Last seen error from the stream, if any.
Tree of all branches for the thread.
Get the metadata for a message, such as first thread state the message
Get tool calls for a specific AI message.
Flattened history of thread states of a thread.
Current interrupt, if the stream is interrupted.
All current interrupts from the stream.
Whether the stream is currently running.
Whether the thread is currently being loaded.
Join an active stream that's already running.
Messages accumulated during the stream.
Set the branch of the thread.
Stops the currently running stream.
Create and stream a run to the thread.
Tool calls paired with their results.
The current state values of the stream.
The ID of the assistant to use.
The current branch of the thread.
LangGraph SDK client used to send requests and receive responses.
Last seen error from the stream, if any.
Tree of all branches for the thread.
Get the metadata for a message, such as first thread state the message
Flattened history of thread states of a thread.
Current interrupt, if the stream is interrupted.
All current interrupts from the stream.
Whether the stream is currently running.
Whether the thread is currently being loaded.
Join an active stream that's already running.
Messages accumulated during the stream.
Set the branch of the thread.
Stops the currently running stream.
Create and stream a run to the thread.
The current state values of the stream.