Tool Manager

​

What Does Tool Manager Do?

• Register tools (functions or Tool objects) with automatic metadata extraction
• Organize tools using tags for easy grouping and discovery
• Retrieve tools by name for dynamic access
• Validate arguments and handle errors before execution
• Execute tools, supporting both sync and async functions
• List tools in multiple formats (OpenAI, LangChain, MCP) for interoperability

​

Key Features

• Flexible Registration: Add tools as functions or Tool objects. Metadata and type hints are automatically extracted.
• Comprehensive Lookup: Retrieve tools by unique name. Get clear errors if a tool is missing.
• Format Conversion: Export tools in OpenAI, LangChain, or MCP format for use in different frameworks.
• Tag-Based Organization: Assign tags to group related tools (e.g., “math”, “text”).
• Validation & Error Handling: Uses Pydantic for argument validation and provides informative error messages.
• Sync & Async Support: Works with both synchronous and asynchronous tool functions.

​

Example: Registering and Using Tools

from universal_mcp.tools import ToolManager

def greet(name: str) -> str:
    """Greets the user by name."""
    return f"Hello, {name}!"

manager = ToolManager()
manager.add_tool(greet)

tool = manager.get_tool("greet")
result = tool.execute({"name": "Alice"})
print(result)  # Output: Hello, Alice!

​

Listing Tools in Different Formats

openai_tools = manager.list_tools(format="openai")
langchain_tools = manager.list_tools(format="langchain")
mcp_tools = manager.list_tools(format="mcp")

​

Organizing Tools with Tags

def add(a: int, b: int) -> int:
    """Adds two numbers."""
    return a + b

manager.add_tool(add, tags=["math", "arithmetic"])
math_tools = manager.list_tools(tags=["math"])

​

Best Practices

• Use clear, unique names for tools
• Write detailed docstrings and use type hints
• Handle errors gracefully in your tool functions
• Use tags to organize related tools

​

Adapters