Within a deployed livekit agent what method call can i use t LiveKit Community #ask-ai

Join Slack

Within a deployed livekit agent, what method call ...

# ask-ai

brief-lock-33347

12/25/2025, 9:01 AM

Within a deployed livekit agent, what method call can i use to continue to add context to my agent responses?

tall-airline-16756

12/25/2025, 9:01 AM

Thanks for reaching out, Hasan. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.

tall-airline-16756

12/25/2025, 9:01 AM

Thanks — you're asking which method to call inside a deployed LiveKit agent to persist additional context for subsequent responses. Use the update_chat_ctx method on the agent instance. Call it after you modify the agent's ChatContext (for example in an on_user_turn_completed hook) so the added messages or data are saved into the chat history and available for future LLM generations. If you’d like, tell me whether you’re using Python or Node.js and I’ll show a short example. Sources: Pipeline nodes and hooks | LiveKit docs | External data and RAG | LiveKit docs

brief-lock-33347

12/25/2025, 9:03 AM

i am uding python

tall-airline-16756

12/25/2025, 9:03 AM

Thanks — since you’re using Python, here’s a concise example showing how to add context to the agent’s

ChatContext

and persist it with

update_chat_ctx

Copy code

# inside your agent class

async def on_user_turn_completed(self, turn_ctx: ChatContext, new_message: ChatMessage) -> None:
    # 1) Retrieve or build additional context (RAG result, DB lookup, etc.)
    rag_content = await my_rag_lookup(new_message.text_content)

    # 2) Add that content to the chat context so it becomes part of history
    turn_ctx.add_message(
        role="assistant",
        content=f"Additional context: {rag_content}"
    )

    # 3) Persist the updated ChatContext so future LLM generations see it
    await self.update_chat_ctx(turn_ctx)

Notes: • Place this in your agent handler (e.g., inside

on_user_turn_completed

) after you compute any extra context. •

turn_ctx.add_message(...)

can also use other roles and structured content if needed. •

update_chat_ctx

ensures the added message is stored for subsequent turns. Want me to adapt this to include metadata, multiple messages, or a specific RAG pattern?

brief-lock-33347

12/25/2025, 9:07 AM

assume i want to add it on init

tall-airline-16756

12/25/2025, 9:07 AM

Got it — you want to persist initial context when the agent starts. Add messages to a ChatContext during initialization and call

update_chat_ctx

to save them. Example for a Python agent:

Copy code

class MyAgent(Agent):
    async def on_init(self) -> None:
        # create a ChatContext (or load an existing one)
        initial_ctx = ChatContext()

        # add one or more messages you want persisted
        initial_ctx.add_message(role="system", content="You are a helpful assistant for Acme Corp.")
        initial_ctx.add_message(role="assistant", content="Here is some product knowledge: ...")

        # persist the chat context so future turns include this context
        await self.update_chat_ctx(initial_ctx)

Notes: • Use

role="system"

role="assistant"

, or other roles as appropriate. • You can load RAG results or DB content here before calling

update_chat_ctx

. • If your agent framework provides a pre-existing ChatContext on init, modify that instance instead of creating a new one.

brief-lock-33347

12/25/2025, 9:09 AM

where do i import chat context from?

tall-airline-16756

12/25/2025, 9:09 AM

Import

ChatContext

from

livekit.agents.llm

Copy code

from livekit.agents.llm import ChatContext

Would you like a brief example showing

on_init

using that import? Sources: External data and RAG | LiveKit docs

3 Views

Open in Slack

Previous Next