Overview

MCP Prompts are an optional part of the Model Context Protocol that let a server expose reusable prompt presets to clients. Clients can list available prompts and request a specific prompt (optionally with arguments) to prefill conversations or tool calls.

Rapid MCP generally does not require prompts because tools already encode strong structure and context. However, some clients expect or benefit from MCP Prompts to provide quick-start prompts, workflows, or standardized system instructions.

TLDR;

Use this if you want to format consistent prompt (with arguments) to the model.

When to use MCP Prompts

• Standardize starter instructions across clients (e.g., a consistent system prompt)
• Offer preset workflows (e.g., "Summarize a document", "Draft an email")
• Parameterize prompts with arguments (e.g., tone, audience) that clients can fill
• Provide discoverability via a list of named prompts

If your agent or UI already manages its own presets, you may not need MCP Prompts. They’re there for compatibility and convenience.

MCP Protocol: Prompts

prompts/list

List the available prompts and their argument descriptors.

Request:

{
  "jsonrpc": "2.0",
  "id": 1,
  "method": "prompts/list",
  "params": {}
}

Response:

{
  "jsonrpc": "2.0",
  "id": 1,
  "result": {
    "prompts": [
      {
        "id": "summarize",
        "name": "Summarize Text",
        "description": "Summarize a passage with an optional target length and tone.",
        "arguments": [
          { "name": "length", "description": "short | medium | long", "required": false },
          { "name": "tone", "description": "professional | friendly | neutral", "required": false }
        ]
      },
      {
        "id": "write-email",
        "name": "Write Email",
        "description": "Draft an email with a topic and audience.",
        "arguments": [
          { "name": "topic", "description": "What the email is about", "required": true },
          { "name": "audience", "description": "Who will receive the email", "required": false }
        ]
      }
    ]
  }
}

prompts/get

Retrieve a specific prompt, optionally templated with arguments. The result provides messages (e.g., system/user) the client can inject into a conversation.

Request:

{
  "jsonrpc": "2.0",
  "id": 2,
  "method": "prompts/get",
  "params": {
    "id": "summarize",
    "arguments": { "length": "short", "tone": "professional" }
  }
}

Response:

{
  "jsonrpc": "2.0",
  "id": 2,
  "result": {
    "messages": [
      {
        "role": "system",
        "content": [ { "type": "text", "text": "You are a helpful assistant. Produce a concise, professional summary." } ]
      },
      {
        "role": "user",
        "content": [ { "type": "text", "text": "Please summarize the following text in a short form with a professional tone." } ]
      }
    ]
  }
}

Notes:

• Servers may template prompt content using provided arguments.
• The exact message shape follows MCP content structures (text, possibly other content types).

Rapid MCP specifics

• Prompts are optional. Rapid MCP focuses on tools and structured arguments; many use cases won’t need prompts.
• If a client expects prompts, you can enable a minimal set to provide starter flows.
• Prompts do not replace tool definitions; they complement them.