Is your feature request related to a problem? Please describe.

When teaching other folks about MCP, the Python SDK makes it pretty easy to define a tool in which the returned text is the response. However, in the TS SDK, you have to wrap the response in a content object, revealing more details about the protocol and raising questions (what's it mean to have more than one content object, etc.).

It would be great if we could allow a tool callback to simply return text and the invoking function can do the necessary wrapping for the protocol.

Example:

Current

server.tool(
  "add-two-numbers",
  "Add two numbers",
  { 
    a: z.number().describe("The first number to add"), 
    b: z.number().describe("The second number to add") 
  },
  ({ a, b }) => {
    return {
      content: [{
        type: "text",
        text: a + b,
      }]
    }
  }
)

Proposal

server.tool(
  "add-two-numbers",
  "Add two numbers",
  { 
    a: z.number().describe("The first number to add"), 
    b: z.number().describe("The second number to add") 
  },
  ({ a, b }) => a + b,
);