peg-memory

peg-memory is a Node.js library for peg-based memorization. It ships with a default 1-100 peg list, lets people import custom lists, generates vivid memory scenes, and can create real local AI pictures when connected to a local image model server.

Install

npm install peg-memory

Use

import { list, importCustom, imager, saveImage } from "peg-memory";

const pegs = list();
console.log(pegs[28]); // Knife

const result = await imager(28, "American Revolution");

console.log(result.story);
console.log(result.explanation);

if (result.image) {
  await saveImage(result, "./revolution.png");
}

importCustom({
  1: "Sun",
  2: "Shoe",
  3: "Tree"
});

API

list()

Returns the active peg list.

defaultList()

Returns the built-in default peg list.

importCustom(customList, options?)

Imports a custom peg list.

• customList: object keyed from 1 to 100
• options.merge: defaults to true

When merge is true, missing pegs fall back to the default list.

resetList()

Restores the built-in default list.

getPeg(number)

Gets one peg by number.

imager(number, thing, options?)

Creates a mnemonic result for a peg and a thing to remember.

Returns:

{
  number: 28,
  peg: "Knife",
  thing: "American Revolution",
  story: "...",
  imagePrompt: "...",
  explanation: "...",
  provider: "fallback",
  svg: "<svg ...",
  dataUri: "data:image/svg+xml..."
}

The svg and dataUri fields give you a real generated image card even in fallback mode.

Text provider options:

• fallback: always works locally with no AI setup
• auto: tries Ollama, then OpenAI, then fallback
• ollama: uses a local Ollama server
• openai: uses the OpenAI API

By default, imager() also tries to use a local image server at http://127.0.0.1:7860. If none is running, it falls back without failing.

Image options:

• image.enabled: turns on real image generation
• image.provider: automatic1111, forge, or none
• image.endpoint: defaults to http://127.0.0.1:7860
• image.model: optional checkpoint/model name
• image.width, image.height, image.steps, image.cfgScale, image.samplerName
• image.promptPrefix: optional style prefix added to the generated prompt
• image.negativePrompt: optional negative prompt

Example with Ollama:

const result = await imager(42, "photosynthesis", {
  provider: "ollama",
  model: "llama3.1:8b"
});

Example with OpenAI:

const result = await imager(42, "photosynthesis", {
  provider: "openai",
  apiKey: process.env.OPENAI_API_KEY
});

Example with a local Stable Diffusion WebUI or Forge server:

const result = await imager(42, "photosynthesis", {
  image: {
    provider: "automatic1111",
    endpoint: "http://127.0.0.1:7860",
    width: 1024,
    height: 1024,
    promptPrefix: "surreal educational illustration, highly detailed"
  }
});

console.log(result.image.dataUri);
await saveImage(result, "./photosynthesis.png");

imagers.batch(entries, options?)

Creates multiple mnemonic images at once.

const results = await imagers.batch([
  { number: 5, thing: "gravity" },
  { number: 12, thing: "mitosis" }
]);

Notes

• The built-in fallback mode is dependency-free and works out of the box.
• Local text AI and local image AI are optional.
• For real local pictures, run a compatible local server such as Stable Diffusion WebUI or Forge with API access enabled.
• The package targets Node.js 18+.