Prompts

What it does

Prompts allow servers to expose parameterized prompt templates that clients can discover and use. A prompt has a name, description, and optional arguments. When a client calls prompts/get, the server returns a fully-rendered message sequence. This enables servers to provide domain-specific prompt engineering that the LLM can leverage without the client needing to craft prompts from scratch.

How it works

Example implementation

server.prompt("code_review", { language: z.string(), code: z.string() },
  async ({ language, code }) => ({
    messages: [{
      role: "user",
      content: { type: "text", text: `Review this ${language} code:\n${code}` }
    }]
  })
);