Outsource MCP

Instructions

Delegate text generation to another AI model. Use this when you need capabilities
or perspectives from a different model than yourself.

Args:
    provider: The AI provider to use (e.g., "openai", "anthropic", "google", "groq")
    model: The specific model identifier (e.g., "gpt-4o", "claude-3-5-sonnet-20241022", "gemini-2.0-flash-exp")
    prompt: The instruction or query to send to the external model

Returns:
    The text response from the external model, or an error message if the request fails

Example usage:
    To get a different perspective: provider="anthropic", model="claude-3-5-sonnet-20241022", prompt="Analyze this problem from a different angle..."
    To leverage specialized models: provider="deepseek", model="deepseek-coder", prompt="Write optimized Python code for..."

Implementation Reference

• server.py:64-108 (handler)

  The core handler function for the 'outsource_text' tool. Decorated with @mcp.tool() for automatic registration and schema inference from type hints and docstring. It maps the provider to a model class, creates an Agent, and executes the prompt to return the generated text.

  @mcp.tool()
  async def outsource_text(provider: str, model: str, prompt: str) -> str:
      """
      Delegate text generation to another AI model. Use this when you need capabilities
      or perspectives from a different model than yourself.
  
      Args:
          provider: The AI provider to use (e.g., "openai", "anthropic", "google", "groq")
          model: The specific model identifier (e.g., "gpt-4o", "claude-3-5-sonnet-20241022", "gemini-2.0-flash-exp")
          prompt: The instruction or query to send to the external model
  
      Returns:
          The text response from the external model, or an error message if the request fails
  
      Example usage:
          To get a different perspective: provider="anthropic", model="claude-3-5-sonnet-20241022", prompt="Analyze this problem from a different angle..."
          To leverage specialized models: provider="deepseek", model="deepseek-coder", prompt="Write optimized Python code for..."
      """
      try:
          # Get the appropriate model class based on provider
          provider_lower = provider.lower()
  
          if provider_lower not in PROVIDER_MODEL_MAP:
              raise ValueError(f"Unknown provider: {provider}")
  
          model_class = PROVIDER_MODEL_MAP[provider_lower]
  
          # Create the agent
          agent = Agent(
              model=model_class(id=model),
              name="Text Generation Agent",
              instructions="You are a helpful AI assistant. Respond to the user's prompt directly and concisely.",
          )
  
          # Run the agent and get response
          response = await agent.arun(prompt)
  
          # Extract the text content from the response
          if hasattr(response, "content"):
              return response.content
          else:
              return str(response)
  
      except Exception as e:
          return f"Error generating text: {str(e)}"

• server.py:32-61 (helper)

  Global dictionary mapping lowercase provider names to their corresponding model classes from the 'agno' library, used within the outsource_text handler to dynamically instantiate the correct model based on user input.

  # Provider to model class mapping
  PROVIDER_MODEL_MAP = {
      "openai": OpenAIChat,
      "anthropic": Claude,
      "google": Gemini,
      "groq": Groq,
      "deepseek": DeepSeek,
      "xai": xAI,
      "perplexity": Perplexity,
      "cohere": Cohere,
      "fireworks": Fireworks,
      "huggingface": HuggingFace,
      "mistral": MistralChat,
      "nvidia": Nvidia,
      "ollama": Ollama,
      "openrouter": OpenRouter,
      "sambanova": Sambanova,
      "together": Together,
      "litellm": LiteLLM,
      "vercel": v0,
      "v0": v0,
      "aws": AwsBedrock,
      "bedrock": AwsBedrock,
      "azure": AzureAIFoundry,
      "cerebras": Cerebras,
      "meta": Llama,
      "deepinfra": DeepInfra,
      "ibm": WatsonX,
      "watsonx": WatsonX,
  }

• server.py:64-64 (registration)

  The @mcp.tool() decorator registers the outsource_text function as an MCP tool, automatically generating schema from signature and docstring.

  @mcp.tool()