plot

This MCP server implements a context-aware storytelling assistant built with Node.js and Express. It uses a structured MCP context (system, user, steps, resources) to guide the generation of story plots via OpenAI. You can fetch the current context at /context, generate a plot by posting to /step/plot, and extend the context by adding characters through /update/characters. The plot generation respects the defined goals and constraints in context.json, and characters can be added dynamically to influence future plots. This setup demonstrates how altering the MCP context changes the LLM's output in a reproducible workflow.

Prerequisites:

• Node.js v18+ installed
• npm installed
• An OpenAI API key

1. Clone the repository and install dependencies

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

2. Create a .env file with your OpenAI API key

OPENAI_API_KEY=sk-xxxxxxxxxxxxxxxxxxxxxxxx

3. Create or customize context.json as needed (example provided in README)

4. Start the server

node server.js

5. Interact with the MCP endpoints:

• Get context: curl http://localhost:3000/context
• Generate a plot: curl -X POST http://localhost:3000/step/plot
• Add a character: curl -X POST http://localhost:3000/update/characters -H "Content-Type: application/json" -d '{"name":"ユキ","role":"謎の旅人","trait":"静かで何かを知っているような雰囲気"}'

Notes:

• Ensure OPENAI_API_KEY is valid and has access to GPT-4 or GPT-4-turbo for best results.
• The context.json file defines the initial state; adjust system/user/steps/resources to fit your project.
• The MCP flow uses the /step/plot endpoint to generate a plot based on the current context; subsequent plots will reflect changes in characters or world settings.
• If the server cannot reach OpenAI, verify network access and key configuration. Ensure the environment variable is correctly loaded (dotenv in this setup).