@cubie/plugin-jupiter

This package is part of the Cubie plugin repository for the (Maiar)[https://maiar.dev] ecosystem, designed to work seamlessly with @maiar-ai/core.

Documentation

For detailed documentation, examples, and API reference, visit: https://maiar.dev/docs

Configuration

The Jupiter plugin requires the following configuration

interface PluginJupiterConfig {
  apiKey?: string; // Your jupiter x-api-key
  store?: (tokens: JupiterTokenResponse[]) => Promise<void>; // An optional method for storing the remote token list to a internal database/store
  load?: (params: LoadJupiterTokenParams) => Promise<JupiterTokenResponse[]>; // An optional method to load and search the synced token list by token name and symbol
}

Required Configuration

All the configurations fields are optional

Optional Configuration

• apiKey: This is your Jupiter x-api-key. Optional (default: '')
• store: A function that accepts a list of JupiterTokenResponse objects and stores them in a database or filesystem
• loader: A function that accepts an object {name?: string, symbol?: string} that can be used to search the tokens stored by store()

Plugin Information

Actions

• get_contract_address: Get the contract address of a token when provided the token name or symbol. This method requires that JupiterService was provided with store and loader methods
• get_quote: Fetch a quote between 2 input mints when provided with an amount. The swapDirection in this method may be inferred due to the nature of the users request.
• get_price: Get the price of a list of input mints with an optional vsToken that can be supplied to use as the base of the price (default: usd).
• get_token_info: Get standard token info when supplied with a contract / mint address.

For more detailed examples and advanced usage, visit our documentation.

Examples

Creation

First you create a storage/loader functions for the token information. In this example we are mocking up a memory store.

const memoryStore: JupiterTokenResponse[] = [];

// mock a storage function to store data
async function storeData(data: JupiterTokenResponse[]) {
  memoryStory.push(...data);
}

// mock a loader function from an external storage source
async function loadData(params: LoadJupiterTokenParams) {
  return (
    params &&
    memoryStory.filter(
      (token) =>
        (params.name &&
          token.name.toLocaleLowerCase() === params.name.toLocaleLowerCase()) ||
        (params.ticker &&
          token.symbol.toLocaleLowerCase() ===
            params.ticker.toLocaleLowerCase())
    )
  );
}

Then you want to initialize your agent runtime, and pass the storeData() and loadData() methods into the new PluginJupiter() call.

import { createRuntime } from "@maiar-ai/core";

// Import providers
import { OpenAIProvider } from "@maiar-ai/model-openai";
import { SQLiteProvider } from "@maiar-ai/memory-sqlite";

// Import all plugins
import { PluginTextGeneration } from "@maiar-ai/plugin-text";
import { PluginTime } from "@maiar-ai/plugin-time";
import { PluginCharacter } from "@maiar-ai/plugin-character";
// Create and start the agent
const runtime = createRuntime({
  model: new OpenAIProvider({
    model: "gpt-4o",
    apiKey: process.env.OPENAI_API_KEY as string,
  }),
  memory: new SQLiteProvider({
    dbPath: path.join(process.cwd(), "data", "conversations.db"),
  }),
  plugins: [
    new PluginTextGeneration(),
    new PluginTime(),
    new PluginCharacter({
      character: fs.readFileSync(
        path.join(process.cwd(), "character.xml"),
        "utf-8"
      ),
    }),
    new PluginJupiter({
      store: storeData,
      load: loadData,
    }),
  ],
});

Now you can start your agents and test it out.

// Start the runtime if this file is run directly
if (require.main === module) {
  console.log("Starting agent...");
  runtime.start().catch((error) => {
    console.error("Failed to start agent:", error);
    process.exit(1);
  });

  // Handle shutdown gracefully
  process.on("SIGINT", async () => {
    console.log("Shutting down agent...");
    await runtime.stop();
    process.exit(0);
  });
}

Demo

Video

demo.mp4

Running the demo

Start by creating an .env file with 2 properties

Configuration

OPEN_API_KEY=
TELEGRAM_BOT_TOKEN=

Installation

Next you need to install the dependencies:

pnpm install

Running the project

Run the project with

pnpm start