mcp-use/dist/examples/chat_example.js

A utility library for integrating Model Context Protocol (MCP) with LangChain, Zod, and related tools. Provides helpers for schema conversion, event streaming, and SDK usage.

/**
 * Simple chat example using MCPAgent with built-in conversation memory.
 *
 * This example demonstrates how to use the MCPAgent with its built-in
 * conversation history capabilities for better contextual interactions.
 *
 * Special thanks to https://github.com/microsoft/playwright-mcp for the server.
 */
import readline from 'node:readline';
import { ChatOpenAI } from '@langchain/openai';
import { config } from 'dotenv';
import { MCPAgent, MCPClient } from '../index.js';
// Load environment variables from .env file
config();
async function runMemoryChat() {
    // Config file path - change this to your config file
    const config = {
        mcpServers: {
            airbnb: {
                command: 'npx',
                args: ['-y', '@openbnb/mcp-server-airbnb', '--ignore-robots-txt'],
            },
        },
    };
    console.error('Initializing chat...');
    // Create MCP client and agent with memory enabled
    const client = new MCPClient(config);
    const llm = new ChatOpenAI({ model: 'gpt-4o-mini' });
    // Create agent with memory_enabled=true
    const agent = new MCPAgent({
        llm,
        client,
        maxSteps: 15,
        memoryEnabled: true, // Enable built-in conversation memory
    });
    console.error('\n===== Interactive MCP Chat =====');
    console.error('Type \'exit\' or \'quit\' to end the conversation');
    console.error('Type \'clear\' to clear conversation history');
    console.error('==================================\n');
    // Create readline interface for user input
    const rl = readline.createInterface({
        input: process.stdin,
        output: process.stdout,
    });
    const question = (prompt) => {
        return new Promise((resolve) => {
            rl.question(prompt, resolve);
        });
    };
    try {
        // Main chat loop
        while (true) {
            // Get user input
            const userInput = await question('\nYou: ');
            // Check for exit command
            if (userInput.toLowerCase() === 'exit' || userInput.toLowerCase() === 'quit') {
                console.error('Ending conversation...');
                break;
            }
            // Check for clear history command
            if (userInput.toLowerCase() === 'clear') {
                agent.clearConversationHistory();
                console.error('Conversation history cleared.');
                continue;
            }
            // Get response from agent
            process.stdout.write('\nAssistant: ');
            try {
                // Run the agent with the user input (memory handling is automatic)
                const response = await agent.run(userInput);
                console.error(response);
            }
            catch (error) {
                console.error(`\nError: ${error}`);
            }
        }
    }
    finally {
        // Clean up
        rl.close();
        await client.closeAllSessions();
    }
}
if (import.meta.url === `file://${process.argv[1]}`) {
    runMemoryChat().catch(console.error);
}