Skip to main content
Applications are the foundation for integrating with external services and platforms (like GitHub, Google, Slack, etc.) in the Universal MCP system. They expose a set of Actions and Triggers for AI agents to use, enabling seamless automation and real-world interaction.

​
What is an Application?

An Application represents an integration with an external service or platform. Applications provide a structured way to:
  • Manage authentication and credentials
  • Handle API communication
  • Define available actions (tools)
  • Process responses and errors
Applications are the building blocks for connecting your AI agents to the outside world.

​
Base Classes for Applications

The system provides three main base classes for building applications:
  • BaseApplication: The abstract base class that defines the common interface
  • APIApplication: For applications that communicate via HTTP APIs
  • GraphQLApplication: For applications that use GraphQL APIs
See also: Actions and Triggers | Integration and Connection

​
Implementation Example: Building an API Application

Here’s an example of building a simple API Application that fetches a random quote: 
from universal_mcp.applications import APIApplication

class ZenquotesApp(APIApplication):
    def __init__(self, **kwargs) -> None:
        super().__init__(name="zenquotes", **kwargs)

    def get_quote(self) -> str:
        """
        Fetches a random inspirational quote from the Zen Quotes API.

        Returns:
            A formatted string containing the quote and its author in the format 'quote - author'
        """
        url = "https://zenquotes.io/api/random"
        response = self._get(url)
        data = response.json()
        quote_data = data[0]
        return f"{quote_data['q']} - {quote_data['a']}"

    def list_tools(self):
        return [self.get_quote]

​
How to Create a New Application

To create a new application, follow these steps:
  1. Create a new package following the naming convention: universal_mcp_<app_name>
  2. Implement your application class inheriting from one of the base classes
  3. Name your class following the convention: <AppName>App
  4. Define your actions as methods and register them in list_tools()
Example skeleton: 
from universal_mcp.applications import APIApplication

class MyApp(APIApplication):
    def __init__(self, name: str, integration=None, **kwargs):
        super().__init__(name, integration, **kwargs)
        self.base_url = "https://api.example.com"

    def list_tools(self):
        return [self.my_action]

    def my_action(self):
        # Implementation here
        pass

​
Best Practices

  • Use clear, descriptive names for your application and actions.
  • Keep authentication and credential management secure and up-to-date.
  • Document each action and its expected input/output.
  • Handle errors gracefully and provide meaningful error messages.
  • Regularly review and update your application to support new features and maintain compatibility.
Need help? Contact support or check the full documentation.