Plot MCP Server

🚀 Simple MCP Server with Node.js

This is a simple prototype of a Model Context Protocol (MCP) server built using Node.js and Express. It interacts with the OpenAI API to simulate a context-aware support assistant, for example, to assist in writing stories.

🚀 Quick Start

This project is a simple prototype of a Model Context Protocol (MCP) server, which interacts with the OpenAI API to simulate a context-aware support assistant for tasks like story writing.

✨ Features

• Structure your context using system, user, steps, and resources.
• Call the OpenAI API to generate context-based story outlines.
• Dynamically add characters via the API.
• Observe the impact of changing the context on LLM responses.

📦 Installation

Requirements

• Node.js v18+
• npm
• OpenAI API key (Get one)

Steps

1. Clone the repository

git clone https://github.com/takehisa10098/plot-mcp-server.git
cd plot-mcp-server

2. Install dependencies

npm install

3. Create a .env file

OPENAI_API_KEY=sk-xxxxxxxxxxxxxxxxxxxxxxxx

4. Create a context.json file

{
  "system": "あなたは小説執筆のサポートAIです。",
  "user": {
    "goal": "和風ファンタジーの短編小説を書きたい",
    "constraints": ["文字数は3000字以内", "テーマは『喪失と再生』"]
  },
  "steps": [
    { "name": "キャラクター作成", "status": "done" },
    { "name": "プロット生成", "status": "in_progress" }
  ],
  "resources": {
    "characters": [
      { "name": "アカネ", "role": "主人公", "trait": "寡黙で芯が強い" }
    ],
    "world": "戦国時代風の異世界。陰陽術が存在する。"
  }
}

5. Create a server.js file

const express = require('express');
const fs = require('fs');
const dotenv = require('dotenv');
const openai = require('openai');

const app = express();
app.use(express.json());

const port = 3000;
app.listen(port, () => {
    console.log(`Server is running at http://localhost:${port}`);
});

// Define routes
app.get('/steps', (req, res) => {
    // Handle step logic
});

app.post('/update/characters', (req, res) => {
    const character = req.body;
    // Add character to resources
});

6. Create an openai.js configuration file

const OpenAI = require('openai');

const openaiConfig = new OpenAI({
    apiKey: process.env.OPENAI_API_KEY,
    dangerouslyAllowBrowser: true
});

module.exports = openaiConfig;

📚 Documentation

What is MCP?

The Model Context Protocol (MCP) is a format for structuring LLM context information. It allows applications to define system, user, steps, and resources to represent the current state of a conversation or project.

This project demonstrates how to design LLM interactions by modifying the structured context rather than rewriting the prompt each time.

🔗 Related Links

• Anthropic MCP Documentation
• OpenAI Function Calling