Prompts | muppet

You can create reusable prompt templates, this not only allows you to reuse the same prompt in multiple places, but also thanks to mcps you can make then dynamic and share them with others. For eg, you can predefine prompts like "Explain {topic} to me like I'm five", here {topic} is a variable that can be replaced with any value.

You can learn more about prompts in the MCP documentation.

import { Hono } from "hono";
import { describePrompt, mValidator, type PromptResponseType } from "muppet";
import z from "zod";
 
const app = new Hono();
 
app.post(
  "/explain-like-im-5",
  describePrompt({
    name: "Explain like I'm 5",
    description: "A prompt to explain an advance topic to a 5 year old",
    completion: ({ name, value }) => [
      "quantum physics",
      "machine learning",
      "natural language processing",
      "artificial intelligence",
    ],
  }),
  mValidator(
    "json",
    z.object({
      topic: z.string(),
    }),
  ),
  (c) => {
    const { topic } = c.req.valid("json");
    return c.json<PromptResponseType>([
      {
        role: "user",
        content: {
          type: "text",
          text: `Explain ${topic} to me like I'm five`,
        },
      },
    ]);
  },
);

mValidator supports Standard Schema, which means you can use validation libs which supports Standard Schema like zod, valibot, arktype, typebox, etc.

You can define completions using completion prop for your prompts. This allows you to provide the LLM with a list of possible values for the variable in the prompt. For example, if you have a prompt that asks for a city name, you can provide a list of cities as completions.