State Streaming
Stream partial agent state updates to the UI while a tool call is still running.
What is this?
By default, agent state only updates between backend checkpoints, so a long-running tool call, such as writing a full document or drafting an email, appears to the UI as one big burst at the end. For agent-native apps, that feels broken: users expect to watch the output materialize.
State streaming copies the value of one tool argument into an agent state key while the model writes it. The UI, subscribed through useAgent, re-renders every token.
When should I use this?
Use state streaming whenever a tool's output is long-form text or a growing structured value and you want the user to see it assemble in real time. Common shapes:
- A collaborative writing agent that emits a document
- A research agent that accumulates a list of findings
- A planning agent that builds up a step-by-step plan
Without streaming, the user stares at a spinner. With streaming, they see the answer grow token-by-token.
The backend: one streaming state mapping
Install the ADK + AG-UI bridge
pip install ag-ui-adkDeclare the predicted state mapping
ADK state streaming uses PredictStateMapping to map the streaming
write_document tool argument into state["document"]. Add
AGUIToolset() to the agent so CopilotKit can forward the state deltas to
the UI.
from __future__ import annotations
from ag_ui_adk import AGUIToolset
from ag_ui_adk.config import PredictStateMapping
from google.adk.agents import LlmAgent
from google.adk.tools import ToolContext
from agents.shared_chat import get_model, stop_on_terminal_text
def write_document(tool_context: ToolContext, document: str) -> dict:
"""Write a document into shared state.
Whenever the user asks you to write or draft anything (essay, poem,
email, summary, etc.), call this tool with the full content as a
single string. The UI renders state["document"] live as you type.
Argument name `document` mirrors langgraph-python's `write_document`
signature so the shared D5 fixture (`tool_argument="document"`) and
the LGP-aligned PredictStateMapping below stay in lock-step.
"""
tool_context.state["document"] = document
return {"status": "ok", "length": len(document)}
_INSTRUCTION = (
"You are a collaborative writing assistant. Whenever the user asks "
"you to write, draft, or revise any piece of text, ALWAYS call the "
"`write_document` tool with the full content as a single string. "
"Never paste the document into a chat message directly — the document "
"belongs in shared state and the UI renders it live as you type."
)
shared_state_streaming_agent = LlmAgent(
name="SharedStateStreamingAgent",
model=get_model(),
instruction=_INSTRUCTION,
tools=[write_document, AGUIToolset()],
after_model_callback=stop_on_terminal_text,
)
SHARED_STATE_STREAMING_PREDICT_STATE = [
PredictStateMapping(
state_key="document",
tool="write_document",
tool_argument="document",
emit_confirm_tool=False,
stream_tool_call=True,
),
]Pass the mapping to ADKAgent
The mapping does nothing until it reaches the ADKAgent middleware. Pass
it as predict_state, and turn on streaming_function_call_arguments so
the middleware receives the tool arguments while the model writes them.
The showcase server builds each endpoint this way:
from ag_ui_adk import ADKAgent, add_adk_fastapi_endpoint
middleware = ADKAgent(
adk_agent=shared_state_streaming_agent,
user_id="demo_user",
session_timeout_seconds=3600,
use_in_memory_services=True,
predict_state=SHARED_STATE_STREAMING_PREDICT_STATE,
streaming_function_call_arguments=True,
)
add_adk_fastapi_endpoint(app, middleware, path="/shared-state-streaming")With predict_state, the frontend gets document while the model writes
the document argument of write_document. When the tool runs,
tool_context.state stores the final value.
Per-token streaming needs google-adk 1.24.0 or later through Vertex AI.
On older versions, or through Gemini in Google AI Studio, the middleware
logs a UserWarning at startup and streams in larger chunks. The state
still updates, only less often.
Stop after the final model response
The agent above imports stop_on_terminal_text and registers it as its
after_model_callback. This prevents Gemini from issuing the same tool
call again after its final text response. See the canonical
stop_on_terminal_text implementation
when adapting the callback for your project.
The backend pattern is always the same: map one streaming tool argument to one shared-state key. Middleware-backed frameworks usually expose this as a declarative mapping. Direct SDK adapters do the same work in their streaming loop by parsing partial tool arguments and emitting STATE_SNAPSHOT whenever the mapped value changes. When the LLM streams that argument, CopilotKit writes every partial value into shared state before the tool finishes executing. The sample below is the LangGraph (Python) version.
from __future__ import annotations
from ag_ui_adk import AGUIToolset
from ag_ui_adk.config import PredictStateMapping
from google.adk.agents import LlmAgent
from google.adk.tools import ToolContext
from agents.shared_chat import get_model, stop_on_terminal_text
def write_document(tool_context: ToolContext, document: str) -> dict:
"""Write a document into shared state.
Whenever the user asks you to write or draft anything (essay, poem,
email, summary, etc.), call this tool with the full content as a
single string. The UI renders state["document"] live as you type.
Argument name `document` mirrors langgraph-python's `write_document`
signature so the shared D5 fixture (`tool_argument="document"`) and
the LGP-aligned PredictStateMapping below stay in lock-step.
"""
tool_context.state["document"] = document
return {"status": "ok", "length": len(document)}
_INSTRUCTION = (
"You are a collaborative writing assistant. Whenever the user asks "
"you to write, draft, or revise any piece of text, ALWAYS call the "
"`write_document` tool with the full content as a single string. "
"Never paste the document into a chat message directly — the document "
"belongs in shared state and the UI renders it live as you type."
)
shared_state_streaming_agent = LlmAgent(
name="SharedStateStreamingAgent",
model=get_model(),
instruction=_INSTRUCTION,
tools=[write_document, AGUIToolset()],
after_model_callback=stop_on_terminal_text,
)
SHARED_STATE_STREAMING_PREDICT_STATE = [
PredictStateMapping(
state_key="document",
tool="write_document",
tool_argument="document",
emit_confirm_tool=False,
stream_tool_call=True,
),
]
A few things to note:
- The state key must exist in your agent state (
documentin this example). - The tool and argument names must match the exact LLM-facing tool call you want to forward.
- When the tool call completes, its final return value is written to the same key, so the streamed partial eventually becomes the authoritative final value.
The frontend: useAgent + OnStateChanged
The UI side is identical to any other shared-state subscription: useAgent with OnStateChanged gives you a reactive agent.value.state. Add OnRunStatusChanged if you want a live/done indicator.
<script setup lang="ts">
import { computed } from "vue";
import { useAgent, UseAgentUpdate } from "@copilotkit/vue/v2";
type DocumentState = { document?: string };
const { agent } = useAgent({
agentId: "writer",
updates: [
UseAgentUpdate.OnStateChanged,
UseAgentUpdate.OnRunStatusChanged,
],
});
const state = computed(
() => (agent.value?.state ?? {}) as DocumentState,
);
</script>
<template>
<article aria-live="polite">
<span v-if="agent?.isRunning">LIVE</span>
<p>{{ state.document ?? "Waiting for a draft…" }}</p>
</article>
</template>From there, state.document is a string that grows on every token, and agent.isRunning tells you whether to show a streaming indicator.
<script setup lang="ts">
import { CopilotChat, CopilotKitProvider } from "@copilotkit/vue/v2";
import StreamingDocument from "./components/StreamingDocument.vue";
</script>
<template>
<CopilotKitProvider runtime-url="/api/copilotkit">
<StreamingDocument />
<CopilotChat agent-id="writer" />
</CopilotKitProvider>
</template>Ask the writer agent to draft a short document. The LIVE badge shows, and the text grows while the agent writes.
If the runtime does not emit partial state snapshots, the UI sees only the final state. The Vue frontend cannot invent updates the backend never sends.
How is this guide?