mcp-add-prompt

/mcp-add-prompt — Add a Prompt Template

Guide the user through adding a reusable prompt template.

Instructions

Step 1: Understand the Prompt

Ask:

1. What workflow does this prompt support? (code review, debugging, data analysis, etc.)
2. What arguments does it need? Parameters the AI provides when requesting the prompt.
3. What messages should it generate? The conversation structure to return.

Prompts are useful for:

• Common multi-step workflows that combine reading resources and calling tools
• Providing context to the AI about how to use the server effectively
• Standardizing interaction patterns

Step 2: Design Arguments Schema

const argsSchema = z.object({
  topic: z.string().describe("The topic to analyze"),
  depth: z.enum(["brief", "detailed"]).default("brief").describe("Analysis depth"),
});

Step 3: Generate the Prompt

Create src/prompts/<name>.ts:

import { z } from "zod";

export const <name>Args = z.object({
  // ... schema
});

export function <name>Handler(args: z.infer<typeof <name>Args>) {
  return {
    description: "Description of what this prompt does",
    messages: [
      {
        role: "user",
        content: {
          type: "text",
          text: `Your prompt text using ${args.topic}`,
        },
      },
    ],
  };
};

Step 4: Register, Test, Verify

Register in src/index.ts with .prompt(), generate tests, run typecheck + tests.