OWUI-Mult-Agent-Processing/multi-agent-collaboration.py

101 lines
3.6 KiB
Python

"""
title: Multi Agent Collaboration System for Open WebUI
Description: Allows for Multiple Models to act as Agents in collaboration
version: 0.5.6
"""
from pydantic import BaseModel, Field
from fastapi import Request
from typing import Callable, Any
from open_webui.models.users import Users
from open_webui.utils.chat import generate_chat_completion
class EventEmitter:
def __init__(self, event_emitter: Callable[[dict], Any] = None):
self.event_emitter = event_emitter
async def emit(self, description="Unknown state", status="in_progress", done=False):
"""
Send a status event to the event emitter.
:param description: Event description
:param status: Event status
:param done: Whether the event is complete
"""
if self.event_emitter:
await self.event_emitter(
{
"type": "status",
"data": {
"status": status,
"description": description,
"done": done,
},
}
)
class Pipe:
class UserValves(BaseModel):
agent_list: list = (
Field(default=[], description="List of Models to process as agents"),
)
operator_model: str = Field(
default="us.anthropic.claude-3-5-sonnet-20241022-v2:0",
description="Default Operator Model to use",
)
pass
def __init__(self):
pass
async def pipe(
self,
body: dict,
__user__: dict,
__request__: Request,
__event_emitter__: Callable[[dict], Any] = None,
) -> str:
"""
:param __user__: User dictionary containing the user ID
:param __event_emitter__: Optional event emitter for tracking status"
:param body: body dictionary containing the prompt and messages and other settings sent to the LLM(s)
:param __request__: User Request object
:return: The final response from the operator model after processing through all agents
"""
# Initialize Event Emitter
emitter = EventEmitter(__event_emitter__)
# Use the unified endpoint with the updated signature
user = Users.get_user_by_id(__user__["id"])
agents = __user__["valves"].agent_list
operator_model = __user__["valves"].operator_model
number_of_agents = len(agents)
if number_of_agents > 0:
# Process through each agent in the list
for agent_model in agents:
# Temporarily change the model to the agent model
body["model"] = agent_model
print(f"Model being use: {agent_model}")
message = f"Processing request through agent model {agent_model}"
await emitter.emit(
description=message, status="agent_processing", done=True
)
response = await generate_chat_completion(__request__, body, user)
# Add Agent response as context
body["messages"].append(
{
"role": "assistant",
"content": f"{response} \n (Provided by Agent: {agent_model})",
}
)
# set Operator for final processing
body["model"] = operator_model
print(f"Model being use: {operator_model}")
message = f"Final Response from {operator_model}"
await emitter.emit(
description=message, status="final_processing", done=True
)
return await generate_chat_completion(__request__, body, user)