∞ Conversable Agent

Documentation

• Class name: ConversableAgentCreator
• Category: SALT/Language Toolkit/Agents
• Output node: False

This node is designed to facilitate the creation of conversable agents, which are AI entities capable of engaging in dialogue based on a predefined system message. It abstracts the complexities of configuring such agents, allowing users to specify basic parameters like the agent's name and its operational message, optionally integrating a language model for enhanced interaction capabilities.

Input types

Required

• name
  • Specifies the name of the conversable agent to be created, serving as a unique identifier and a way to reference the agent in interactions.
  • Comfy dtype: STRING
  • Python dtype: str
• system_message
  • Defines the initial message or instruction that the agent will use to guide its interactions, setting the tone and scope of its conversational abilities.
  • Comfy dtype: STRING
  • Python dtype: str

Optional

• llm_model
  • An optional parameter that allows for the integration of a language learning model to enhance the agent's conversational capabilities, providing a more dynamic and responsive interaction experience.
  • Comfy dtype: LLM_MODEL
  • Python dtype: dict

Output types

• agent
  • Comfy dtype: AGENT
  • The created conversable agent, ready to be utilized for engaging in dialogues and performing tasks as defined by its configuration.
  • Python dtype: ConversableAgent

Usage tips

• Infra type: CPU
• Common nodes: unknown

Source code

class ConversableAgentCreator:
    @classmethod
    def INPUT_TYPES(cls):
        return {
            "required": {
                "name": ("STRING", {"multiline": False, "placeholder": "Assistant"}),
                "system_message": ("STRING", {
                    "multiline": True,
                    "default": "You are a helpful AI assistant. You can help with document QA. Return 'TERMINATE' when the task is done."
                }),
            },
            "optional": {
                "llm_model": ("LLM_MODEL",),
            }
        }

    RETURN_TYPES = ("AGENT",)
    RETURN_NAMES = ("agent",)

    FUNCTION = "create_agent"
    CATEGORY = f"{MENU_NAME}/{SUB_MENU_NAME}/Agents"

    def create_agent(self, name, system_message, llm_model=None):
        agent = ConversableAgent(
            name=name,
            system_message=system_message,
            llm_config={"config_list": [{"model": llm_model["llm"].model, "api_key": llm_model["llm"].api_key}]} if llm_model is not None else False,
            human_input_mode="NEVER",
        )
        return (agent,)