2254a8d0a2
Smallest UIWorker demo: a voice LLM in the main pipeline delegates screen-relevant utterances to a UIWorker via a respond job; the UIWorker auto-injects the current <ui_state> and answers grounded in what's on screen. Includes a vanilla-JS client that streams accessibility snapshots over RTVI.
275 lines
9.2 KiB
Python
275 lines
9.2 KiB
Python
#
|
|
# Copyright (c) 2026, Daily
|
|
#
|
|
# SPDX-License-Identifier: BSD 2-Clause License
|
|
#
|
|
|
|
"""Hello UIWorker — the smallest possible accessibility-snapshot demo.
|
|
|
|
A voice bot whose LLM delegates every screen-relevant utterance to a
|
|
``UIWorker`` that sees the page and writes the spoken answer.
|
|
|
|
Architecture::
|
|
|
|
Main worker (PipelineWorker, owns transport + RTVI):
|
|
transport.in → STT → user_agg → LLM → TTS → transport.out → assistant_agg
|
|
└── answer_about_screen(query) tool
|
|
└── params.pipeline_worker.job("hello", name="respond", payload={query})
|
|
|
|
HelloWorker (UIWorker):
|
|
└── @tool answer(text)
|
|
|
|
The main LLM is the conversational layer: it forwards every utterance
|
|
to the UI worker via the ``answer_about_screen`` tool and speaks the
|
|
result. The UI worker's built-in ``respond`` job fires, which
|
|
auto-injects the latest ``<ui_state>`` block into its LLM context. The
|
|
UI worker's LLM picks the ``answer`` tool with a spoken reply grounded
|
|
in what's on screen.
|
|
|
|
The RTVI⇄bus UI bridge is built into ``PipelineWorker`` (active because
|
|
``enable_rtvi=True``), so inbound ``ui-snapshot`` messages from the
|
|
client are broadcast on the bus and the ``UIWorker`` stores them — no
|
|
decorator or manual wiring needed.
|
|
|
|
Why two LLMs for "hello world": this is the pattern UIWorker's
|
|
auto-inject is built for. The UI worker auto-injects the current screen
|
|
at the start of every delegated job, so the conversational LLM stays
|
|
small and screen-unaware. Later examples (pointing, form-fill, deixis,
|
|
async-tasks) compose new tools onto the same skeleton.
|
|
|
|
Run::
|
|
|
|
uv run python bot.py
|
|
|
|
Then open the client at ``http://localhost:5173`` (see ``README.md``).
|
|
|
|
Requirements:
|
|
|
|
- OPENAI_API_KEY
|
|
- DEEPGRAM_API_KEY
|
|
- CARTESIA_API_KEY
|
|
"""
|
|
|
|
import os
|
|
|
|
from dotenv import load_dotenv
|
|
from loguru import logger
|
|
|
|
from pipecat.adapters.schemas.tools_schema import ToolsSchema
|
|
from pipecat.audio.vad.silero import SileroVADAnalyzer
|
|
from pipecat.frames.frames import LLMRunFrame
|
|
from pipecat.pipeline.job_context import JobError
|
|
from pipecat.pipeline.pipeline import Pipeline
|
|
from pipecat.pipeline.runner import PipelineRunner
|
|
from pipecat.pipeline.worker import PipelineParams, PipelineWorker
|
|
from pipecat.processors.aggregators.llm_context import LLMContext
|
|
from pipecat.processors.aggregators.llm_response_universal import (
|
|
LLMContextAggregatorPair,
|
|
LLMUserAggregatorParams,
|
|
)
|
|
from pipecat.runner.types import RunnerArguments
|
|
from pipecat.runner.utils import create_transport
|
|
from pipecat.services.cartesia.tts import CartesiaTTSService
|
|
from pipecat.services.deepgram.stt import DeepgramSTTService
|
|
from pipecat.services.llm_service import FunctionCallParams
|
|
from pipecat.services.openai.llm import OpenAILLMService
|
|
from pipecat.transports.base_transport import BaseTransport, TransportParams
|
|
from pipecat.transports.daily.transport import DailyParams
|
|
from pipecat.workers.llm import tool
|
|
from pipecat.workers.ui import UIWorker
|
|
|
|
load_dotenv(override=True)
|
|
|
|
MAIN_NAME = "main"
|
|
|
|
transport_params = {
|
|
"daily": lambda: DailyParams(
|
|
audio_in_enabled=True,
|
|
audio_out_enabled=True,
|
|
),
|
|
"webrtc": lambda: TransportParams(
|
|
audio_in_enabled=True,
|
|
audio_out_enabled=True,
|
|
),
|
|
}
|
|
|
|
|
|
VOICE_PROMPT = """\
|
|
You are the voice layer of a screen-aware assistant. A separate UI \
|
|
layer sees the page the user is looking at and writes the spoken \
|
|
reply for any question that could plausibly involve the page.
|
|
|
|
## Routing rule
|
|
For every user utterance that could involve the page in any way — \
|
|
"what's on screen", "what does this say", "is X on the page", \
|
|
factual questions, navigational questions, anything where the page \
|
|
content might matter — call ``answer_about_screen`` with the user's \
|
|
request verbatim. The tool's response is the spoken reply, already \
|
|
TTS-ready; pass it through without paraphrasing.
|
|
|
|
If the request has nothing to do with the page, still call the \
|
|
tool — the UI layer falls back to general knowledge.
|
|
|
|
## When to answer directly
|
|
Only respond directly for pure pleasantries that don't need any \
|
|
content awareness:
|
|
|
|
- Greetings ("hi", "hello").
|
|
- Acknowledgements ("thanks", "got it").
|
|
- Goodbyes ("bye", "see you").
|
|
|
|
Keep direct replies to one short spoken sentence. No markdown, no \
|
|
lists, no symbols."""
|
|
|
|
|
|
# The UI wire-format guide (UI_STATE_PROMPT_GUIDE) is appended to the LLM's
|
|
# system instruction automatically by UIWorker, so this prompt only needs the
|
|
# app-specific behavior.
|
|
HELLO_PROMPT = """\
|
|
You answer the user's question grounded in the page they're looking \
|
|
at. The current ``<ui_state>`` block is in your context — use it for \
|
|
anything the user could be asking about on screen.
|
|
|
|
Always call exactly one tool: ``answer(text)``. Put the spoken reply \
|
|
in ``text``. Plain language, one or two short sentences, no markdown \
|
|
or symbols.
|
|
|
|
When the question is about something on the page, ground claims in \
|
|
the ``<ui_state>`` content. When it's general knowledge with no \
|
|
on-page referent (history, geography, definitions), answer from your \
|
|
own knowledge. Don't tell the user what you can't see — just answer \
|
|
or admit you don't know."""
|
|
|
|
|
|
class HelloWorker(UIWorker):
|
|
"""Snapshot-aware layer. Answers grounded in ``<ui_state>``.
|
|
|
|
``UIWorker`` defaults to ``active=True`` (unlike ``LLMWorker``)
|
|
because the canonical UIWorker role is an always-on delegate, so it
|
|
is online to receive snapshots and ``respond`` jobs as soon as its
|
|
pipeline starts.
|
|
"""
|
|
|
|
def __init__(self):
|
|
llm = OpenAILLMService(
|
|
api_key=os.environ["OPENAI_API_KEY"],
|
|
settings=OpenAILLMService.Settings(system_instruction=HELLO_PROMPT),
|
|
)
|
|
super().__init__("hello", llm=llm)
|
|
|
|
@tool
|
|
async def answer(self, params: FunctionCallParams, text: str):
|
|
"""Speak ``text`` back to the user.
|
|
|
|
Args:
|
|
text: The spoken reply in plain language. One or two short
|
|
sentences. No markdown, no symbols, no lists.
|
|
"""
|
|
logger.info(f"{self}: answer('{text[:80]}…')")
|
|
await self.respond_to_job(speak=text)
|
|
await params.result_callback(None)
|
|
|
|
|
|
async def answer_about_screen(params: FunctionCallParams, query: str):
|
|
"""Ask the screen-aware UI layer to answer about the current page.
|
|
|
|
Args:
|
|
query (str): The user's request, passed verbatim.
|
|
"""
|
|
logger.info(f"answer_about_screen('{query}')")
|
|
try:
|
|
async with params.pipeline_worker.job(
|
|
"hello", name="respond", payload={"query": query}, timeout=30
|
|
) as t:
|
|
pass
|
|
except JobError as e:
|
|
logger.warning(f"hello job failed: {e}")
|
|
await params.result_callback("Something went wrong on my side.")
|
|
return
|
|
|
|
speak = (t.response or {}).get("speak")
|
|
await params.result_callback(speak or "I'm not sure how to answer that.")
|
|
|
|
|
|
async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
|
|
logger.info("Starting hello-snapshot bot")
|
|
|
|
runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
|
|
|
|
stt = DeepgramSTTService(api_key=os.environ["DEEPGRAM_API_KEY"])
|
|
tts = CartesiaTTSService(
|
|
api_key=os.environ["CARTESIA_API_KEY"],
|
|
settings=CartesiaTTSService.Settings(
|
|
voice=os.getenv("CARTESIA_VOICE_ID", "71a7ad14-091c-4e8e-a314-022ece01c121"),
|
|
),
|
|
)
|
|
llm = OpenAILLMService(
|
|
api_key=os.environ["OPENAI_API_KEY"],
|
|
settings=OpenAILLMService.Settings(system_instruction=VOICE_PROMPT),
|
|
)
|
|
llm.register_direct_function(answer_about_screen, cancel_on_interruption=False, timeout_secs=60)
|
|
|
|
context = LLMContext(tools=ToolsSchema(standard_tools=[answer_about_screen]))
|
|
aggregators = LLMContextAggregatorPair(
|
|
context,
|
|
user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
|
|
)
|
|
|
|
pipeline = Pipeline(
|
|
[
|
|
transport.input(),
|
|
stt,
|
|
aggregators.user(),
|
|
llm,
|
|
tts,
|
|
transport.output(),
|
|
aggregators.assistant(),
|
|
]
|
|
)
|
|
|
|
worker = PipelineWorker(
|
|
pipeline,
|
|
name=MAIN_NAME,
|
|
params=PipelineParams(
|
|
enable_metrics=True,
|
|
enable_usage_metrics=True,
|
|
),
|
|
idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
|
|
)
|
|
|
|
@transport.event_handler("on_client_connected")
|
|
async def on_client_connected(transport, client):
|
|
logger.info("Client connected")
|
|
context.add_message(
|
|
{
|
|
"role": "developer",
|
|
"content": (
|
|
"Greet the user briefly. Tell them they can ask about "
|
|
"anything on this page. One short sentence."
|
|
),
|
|
}
|
|
)
|
|
await worker.queue_frame(LLMRunFrame())
|
|
|
|
@transport.event_handler("on_client_disconnected")
|
|
async def on_client_disconnected(transport, client):
|
|
logger.info("Client disconnected")
|
|
await runner.cancel()
|
|
|
|
await runner.launch_worker(HelloWorker())
|
|
await runner.launch_worker(worker)
|
|
|
|
await runner.run()
|
|
|
|
|
|
async def bot(runner_args: RunnerArguments):
|
|
"""Main bot entry point compatible with Pipecat Cloud."""
|
|
transport = await create_transport(runner_args, transport_params)
|
|
await run_bot(transport, runner_args)
|
|
|
|
|
|
if __name__ == "__main__":
|
|
from pipecat.runner.run import main
|
|
|
|
main()
|