Water: A Versatile Multi-Agent Orchestration Framework

Water is a robust orchestration framework designed to streamline the creation of complex multi-agent systems. This framework enables developers to avoid dependencies on specific agent frameworks, providing flexibility and adaptability. Whether integrating with popular agent frameworks such as LangChain, CrewAI, or Agno, or even utilizing custom-built agents, Water serves as a reliable orchestration layer to coordinate and scale your multi-agent workflows effectively.

Key Features

• Framework Agnostic: Seamlessly integrate any agent framework or custom implementation into your projects.
• Flexible Workflows: Create and manage intricate multi-agent interactions utilizing straightforward Python syntax.
• Playground: Easily run a lightweight FastAPI server to test and execute the workflows defined within the framework.

Example Usage

Here's a basic example demonstrating how to set up a flow that adds five to an input value using Water:

import asyncio  
from water import Flow, create_task  
from pydantic import BaseModel  

class NumberInput(BaseModel):  
    value: int  

class NumberOutput(BaseModel):  
    result: int  

def add_five(params, context):  
    return {"result": params["input_data"]["value"] + 5}  

math_task = create_task(  
    id="math_task",  
    description="Math task",  
    input_schema=NumberInput,  
    output_schema=NumberOutput,  
    execute=add_five  
)  

flow = Flow(id="my_flow", description="My flow").then(math_task).register()  

async def main():  
    result = await flow.run({"value": 10})  
    print(result)  

if __name__ == "__main__":  
    asyncio.run(main())  

Contributing

Contributions from the community are encouraged and greatly appreciated! Ways to contribute include:

• Reporting bugs and requesting features
• Submitting code contributions
• Improving documentation
• Engaging in testing and quality assurance

Roadmap

Future developments for Water include:

• A storage layer for preserving flow sessions and task runs
• Support for human-in-the-loop scenarios
• A retry mechanism for individual tasks

Water is an excellent choice for those looking to build scalable and efficient multi-agent systems with a high degree of flexibility.