gpt

GPT Agent

This example uses the OpenAI API for GPT-based responses and the XMTP protocol for secure messaging. You can test your agent on xmtp.chat or any other XMTP-compatible client.

Environment variables

Add the following keys to a .env file:

WALLET_KEY=    # Private key for XMTP
ENCRYPTION_KEY=         # Secondary key for local encryption
OPENAI_API_KEY=    # e.g., sk-xxx...

Usage

import { Message, xmtpClient } from "@xmtp/agent-starter";
import OpenAI from "openai";

// Initialize OpenAI
const openai = new OpenAI();

async function main() {
  const agent = await xmtpClient({
    walletKey: process.env.WALLET_KEY as string,
    onMessage: async (message: Message) => {
      console.log(
        `Decoded message: ${message?.content.text} from ${message.sender.address}`,
      );

      // Send user text to OpenAI
      const completion = await openai.chat.completions.create({
        model: "gpt-4",
        messages: [
          { role: "system", content: "You are a helpful assistant." },
          { role: "user", content: message?.content.text ?? "" },
        ],
      });

      // Extract GPT response
      const gptMessage = completion.choices[0]?.message?.content?.trim();

      // Send GPT response back via XMTP
      await client.send({
        message: gptMessage ?? "",
        originalMessage: message,
      });
    },
  });

  console.log(
    `XMTP client initialized on ${client.address}\n` +
      `Try sending a message at https://xmtp.chat/dm/${client.address}`,
  );
}

main().catch(console.error);

Run the agent and send a test message from xmtp.chat.
Enjoy your GPT-powered XMTP agent!