If there are multiple tools available, is there a way to prioritize a subset of tools to be always used by the agent?

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

Yes i want to set few tools to be aclled always by the agent

Hey there! Let's talk about how you can customize LiveKit to make your agents even more awesome by prioritizing specific tools. We'll use the super handy

tool_choice

parameter in the

ModelSettings

class, which lets you tell the system exactly which tools you want to shine. Here's a friendly example of how you can set this up:

@dataclass
class ModelSettings:
    tool_choice: NotGivenOr[llm.ToolChoice] = NOT_GIVEN
    """The tool choice to use when calling the LLM."""

And guess what? The

Agent

class is also super flexible. When you're setting up your agent, you can define tool preferences right in the initialization parameters:

def __init__(
    self,
    *,
    instructions: str,
    chat_ctx: NotGivenOr[llm.ChatContext | None] = NOT_GIVEN,
    tools: list[llm.FunctionTool | llm.RawFunctionTool] | None = None,
    # other parameters...
) -> None:

By tweaking these settings, you'll have your agents working exactly the way you want. Sounds cool, right? If you need any help putting this into practice, I'm totally here for you – just give me a shout! Sources: GitHub::LiveKit Agents | livekit-agents/livekit/agents/voice/agent.py | GitHub:: LiveKit Composite Repo | livekit/agents/livekit-agents/livekit/age...

Can you explain with an example

I don't have the answer you're looking for. You could also try asking your question: • in one of the other Slack channels or • to https://deepwiki.com/livekit/livekit_composite which is trained on all LiveKit source code If you find the answer, please post it here to help others!

Hey there! Let's talk about how you can customize LiveKit to make your agents even more awesome by prioritizing specific tools. We'll use the super handy

tool_choice

parameter in the

ModelSettings

class, which lets you tell the system exactly which tools you want to shine. Here's a friendly example of how you can set this up:

@dataclass
class ModelSettings:
    tool_choice: NotGivenOr[llm.ToolChoice] = NOT_GIVEN
    """The tool choice to use when calling the LLM."""

And guess what? The

Agent

class is also super flexible. When you're setting up your agent, you can define tool preferences right in the initialization parameters:

def __init__(
    self,
    *,
    instructions: str,
    chat_ctx: NotGivenOr[llm.ChatContext | None] = NOT_GIVEN,
    tools: list[llm.FunctionTool | llm.RawFunctionTool] | None = None,
    # other parameters...
) -> None:

By tweaking these settings, you'll have your agents working exactly the way you want. Sounds cool, right? If you need any help putting this into practice, I'm totally here for you – just give me a shout! Use this as context