wabee 0.3.2

Wabee AI Software Development Kit

Wabee SDK

wabee-sdk is a Python module for development of modules and extensions for the Wabee agentic AI platform.

Installation

pip install wabee

Command Line Interface (CLI)

The Wabee SDK includes a powerful CLI tool to streamline the development of Wabee agent tools.

Creating a New Tool

Create a new tool using the interactive CLI:

wabee tools create

This command will prompt you for:

• Tool name
• Tool type (simple or complete)
• Tool description
• Initial version

Tool Types

1. Simple Tool:

   • Ideal for straightforward, single-function tools
   • Uses the @simple_tool decorator
   • Less boilerplate code
   • Perfect for quick implementations

2. Complete Tool:

   • Full class implementation
   • More control over tool behavior
   • Better for complex tools with multiple operations
   • Includes error handling infrastructure

Building Tool Containers

Build a tool into a container image:

wabee tools build <tool_directory> [options]

Options:

• --image: Specify custom image name (default: toolname:latest)

Example:

wabee tools build ./my-tool

Tool Project Structure

When you create a new tool, the following structure is generated:

my_tool/
├── my_tool_tool.py      # Main tool implementation
├── requirements.txt     # Python dependencies
├── server.py           # gRPC server implementation
└── toolspec.yaml       # Tool specification and metadata

RPC Server

Each built tool runs as a gRPC server that exposes a standardized interface for tool execution. The server:

• Listens on port 50051 by default (configurable via WABEE_GRPC_PORT)
• Automatically handles input validation using your Pydantic schemas
• Provides standardized error handling and reporting
• Supports streaming responses for long-running operations

When you build a tool with wabee tools build, the resulting container image includes:

• Your tool implementation
• All dependencies
• A pre-configured gRPC server
• Generated protocol buffers for type-safe communication

You can run the built container with:

docker run -p 50051:50051 mytool:latest

toolspec.yaml

The tool specification file contains metadata about your tool:

tool:
  name: MyTool
  description: Your tool description
  version: 0.1.0
  entrypoint: my_tool_tool.py

Requirements

• Python >=3.11,<3.12
• Docker (for building containers)
• Internet connection (for downloading S2I builder)

Development Examples

Simple Tool Example

from pydantic import BaseModel
from wabee.tools.simple_tool import simple_tool

class MyToolInput(BaseModel):
    message: str

@simple_tool(schema=MyToolInput)
async def my_tool(input_data: MyToolInput) -> str:
    return f"Processed: {input_data.message}"

Complete Tool Example

from typing import Optional, Type
from pydantic import BaseModel
from wabee.tools.base_tool import BaseTool
from wabee.tools.tool_error import ToolError

class MyToolInput(BaseModel):
    message: str

class MyTool(BaseTool):
    args_schema: Type[BaseModel] = MyToolInput

    async def execute(self, input_data: MyToolInput) -> tuple[Optional[str], Optional[ToolError]]:
        try:
            result = f"Processed: {input_data.message}"
            return result, None
        except Exception as e:
            return None, ToolError(type="EXECUTION_ERROR", message=str(e))

Contributing

Suggestions are welcome! Please feel free to submit bug reports or feedbacks as a Github issues.

Links

• Website: https://wabee.ai/
• Documentation: https://documentation.wabee.ai
• GitHub: https://github.com/wabee-ai/wabee-sdk