Skip to main content

Integrating AI-Engine with UAgents

Project description

UAgents AI-Engine Integration

📌 Overview

This package provides the necessary types for integrating AI-Engine with UAgents, enabling structured responses and request handling within the UAgents framework. It includes models for handling various response types, key-value pairs, and booking requests.

Installation

To install the package, use the following command:

pip install uagents-ai-engine

Usage

Importing the Package

To use the models provided by this package, import them as follows:

from ai_engine.chitchat import ChitChatDialogue
from ai_engine.messages import DialogueMessage
from ai_engine.dialogue import EdgeMetadata, EdgeDescription

The chitchat, messages, and dialogue modules provide essential classes, types, and functions to facilitate structured and dynamic interactions with the agent. These modules support an open-ended communication model, allowing users to engage in an ongoing dialogue with the agent. After each user message, the agent responds, enabling a continuous and interactive conversation that can proceed as long as needed.

from ai_engine.types import UAgentResponseType, KeyValue, UAgentResponse, BookingRequest

The types module offers a comprehensive set of models for handling responses, key-value pairs, and booking requests. This module is designed for scenarios where a single exchange is sufficient. The user sends a message, receives a well-structured response from the agent, and the interaction concludes efficiently. This approach is ideal for straightforward queries and tasks.

📦 Components

1. Response Models

The following classes are used for non-dialogue agent communication.

UAgentResponseType

An enumeration defining the types of responses that an agent can provide:

class UAgentResponseType(Enum):
    FINAL = "final"
    ERROR = "error"
    VALIDATION_ERROR = "validation_error"
    SELECT_FROM_OPTIONS = "select_from_options"
    FINAL_OPTIONS = "final_options"

KeyValue

A model representing a key-value pair. It is usually used by the AI-Engine as a way to provide multiple choice options to the user:

class KeyValue(Model):
    key: str
    value: str

UAgentResponse

A model for structuring the response from an agent:

class UAgentResponse(Model):
    version: Literal["v1"] = "v1"
    type: UAgentResponseType
    request_id: Optional[str]
    agent_address: Optional[str]
    message: Optional[str]
    options: Optional[List[KeyValue]]
    verbose_message: Optional[str]
    verbose_options: Optional[List[KeyValue]]

Attributes:

  • version: The version of the response model (default is "v1").
  • type: The type of the response, based on UAgentResponseType.
  • request_id: An optional identifier for the request.
  • agent_address: An optional address of the agent.
  • message: An optional message from the agent.
  • options: An optional list of key-value options.
  • verbose_message: An optional verbose message from the agent.
  • verbose_options: An optional list of verbose key-value options.

BookingRequest

A model for handling booking requests:

class BookingRequest(Model):
    request_id: str
    user_response: str
    user_email: str
    user_full_name: str

Attributes:

  • request_id: The unique identifier for the booking request.
  • user_response: The response from the user.
  • user_email: The email address of the user.
  • user_full_name: The full name of the user.

2. Dialogue Management

ChitChatDialogue

A specific dialogue class for AI-Engine enabled chit-chat:

class ChitChatDialogue(Dialogue):
    def on_initiate_session(self, model: Type[Model]):
        # ... (session initiation logic)

    def on_reject_session(self, model: Type[Model]):
        # ... (session rejection logic)

    def on_start_dialogue(self, model: Type[Model]):
        # ... (dialogue start logic)

    def on_continue_dialogue(self, model: Type[Model]):
        # ... (dialogue continuation logic)

    def on_end_session(self, model: Type[Model]):
        # ... (session end logic)

How to initialize a ChitChatDialogue instance:

agent = Agent()

# instantiate the dialogues
chitchat_dialogue = ChitChatDialogue(
    version="0.1",
    storage=agent.storage,
)

For a more in depth example, see the ChitChatDialogue example.

3. Extending Dialogue with Metadata

EdgeMetadata

Metadata for the edges to specify targets and observability:

  • system implies AI Engine processing
  • user is direct message to the user
  • ai is a message to the AI Engine
  • agent is a message to the agent.
class EdgeMetadata(BaseModel):
    target: Literal["user", "ai", "system", "agent"]
    observable: bool

EdgeDescription

A structured description for the edge:

class EdgeDescription(BaseModel):
    description: str
    metadata: EdgeMetadata

Create Edge Function

Function to create an edge with metadata:

init_session = create_edge(
    name="Initiate session",
    description="Every dialogue starts with this transition.",
    target="user",
    observable=True,
    parent=default_state,
    child=init_state,
)

3. Message Types

The following classes are used for dialogue agent communication.

BaseMessage

A base model for all messages:

class BaseMessage(Model):
    message_id: UUID
    timestamp: datetime

DialogueMessage

A model for generic dialogue messages:

class DialogueMessage(BaseMessage):
    type: Literal["agent_message", "agent_json", "user_message"]
    agent_message: Optional[str]
    agent_json: Optional[AgentJSON]
    user_message: Optional[str]

Can be initialized as follows, we'll call this class ChitChatDialogueMessage:

class ChitChatDialogueMessage(DialogueMessage):
    """ChitChat dialogue message"""

    pass

And then use it as follows:

@chitchat_dialogue.on_continue_dialogue(ChitChatDialogueMessage)

Where chitchat_dialogue is defined above in the ChitChatDialogue section and on_continue_dialogue is a method of the ChitChatDialogue class that can be extended.

AI-Engine Integration

This integration adds the required types for AI-Engine to interact with UAgents effectively. The UAgentResponse model serves as the primary structure for agent responses, while BookingRequest handles user booking requests.

Digest

UAgentResponse digest:

model:cf0d1367c5f9ed8a269de559b2fbca4b653693bb8315d47eda146946a168200e

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

uagents_ai_engine-0.6.0.tar.gz (5.8 kB view details)

Uploaded Source

Built Distribution

uagents_ai_engine-0.6.0-py3-none-any.whl (7.6 kB view details)

Uploaded Python 3

File details

Details for the file uagents_ai_engine-0.6.0.tar.gz.

File metadata

  • Download URL: uagents_ai_engine-0.6.0.tar.gz
  • Upload date:
  • Size: 5.8 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/5.0.0 CPython/3.12.2

File hashes

Hashes for uagents_ai_engine-0.6.0.tar.gz
Algorithm Hash digest
SHA256 fb9e85c2db9652b43797ae35e864eeee1502604ad1efd055fe5de4526e2350ed
MD5 b8ed18201e234ad8a228855145c68e8b
BLAKE2b-256 c818c48bb8ae6a6be61ceb3bdf056c3dc3fba7409ab8370046e1aab8193c9ad1

See more details on using hashes here.

File details

Details for the file uagents_ai_engine-0.6.0-py3-none-any.whl.

File metadata

File hashes

Hashes for uagents_ai_engine-0.6.0-py3-none-any.whl
Algorithm Hash digest
SHA256 6162209f5c83cc7b56f1eff323956f3ed1274a393170e4bcc657ba04cfd7b8b2
MD5 2d35fe0fe4b3a4ca04fdf74459948fd1
BLAKE2b-256 c0ec88def2f31afc16c1a53aa1e65eb5f4114880123ebde33f30d913f70c73b3

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page