Skip to main content

​
Configuration Overview

mcp_use configuration is organized into two main areas: Client Configuration for connecting to MCP servers, and Agent Configuration for customizing agent behavior and LLM integration.

​
Configuration Architecture

mcp_use follows a clear separation between client-side and agent-side concerns:

Client Configuration

MCPClient Setup
  • MCP server connections
  • Multi-server configurations
  • Sandboxed execution
  • Connection types

Agent Configuration

MCPAgent Setup
  • API key management
  • LLM integration
  • Server manager
  • Tool access control
  • Memory and prompts
  • Adapter usage

​
Quick Start Configuration

For a basic setup, you need both client and agent configuration:

​
1. Client Setup

import { MCPClient } from 'mcp-use'

// Configure your MCP servers
const config = {
    mcpServers: {
        playwright: {
            command: 'npx',
            args: ['@playwright/mcp@latest'],
            env: { DISPLAY: ':1' }
        }
    }
}

const client = new MCPClient(config)

​
2. Agent Setup

import { MCPAgent } from 'mcp-use'
import { ChatOpenAI } from '@langchain/openai'

// Configure your agent with an LLM
const llm = new ChatOpenAI({ model: 'gpt-4o' })
const agent = new MCPAgent({ llm, client })

​
3. Basic Usage

async function main() {
    const result = await agent.run('Search for information about climate change')
    console.log(result)
}

main().catch(console.error)

​
Configuration Paths

1

Client Configuration

Set up your MCPClient to connect to MCP servers. This includes configuring server connections, managing API keys, and setting up multi-server environments.Start here: Client Configuration Guide →
2

Agent Configuration

Configure your MCPAgent’s behavior, including LLM integration, tool restrictions, memory settings, and custom prompts.Continue with: Agent Configuration Guide →
3

Advanced Topics

Explore connection types, server management, and LLM integration patterns for complex use cases.Learn more: Connection Types | Server Manager | LLM Integration

​
Common Configuration Patterns

​
Development Setup

// Simple development configuration
import { config } from 'dotenv'
import { loadConfigFile, MCPAgent, MCPClient } from 'mcp-use'
import { ChatOpenAI } from '@langchain/openai'

config()

const configuration = await loadConfigFile('dev-config.json')
const client = new MCPClient(configuration)
const agent = new MCPAgent({
    llm: new ChatOpenAI({ model: 'gpt-4o' }),
    client,
    maxSteps: 10,
    verbose: true
})

​
Production Setup

// Production configuration with restrictions
const agent = new MCPAgent({
    llm: new ChatOpenAI({ model: 'gpt-4o', temperature: 0.1 }),
    client,
    maxSteps: 30,
    disallowedTools: ['file_system', 'shell'],
    useServerManager: true,
    memoryEnabled: true
})

​
Multi-Server Setup

// Complex multi-server configuration
const config = await loadConfigFile('multi-server.json')
const client = new MCPClient(config)
const agent = new MCPAgent({
    llm,
    client,
    useServerManager: true,  // Auto-select servers
    systemPrompt: 'You have access to web browsing, file operations, and API tools.'
})

​
What’s Next?

Client Configuration

Learn how to configure MCPClient and connect to MCP servers

Agent Configuration

Discover how to customize MCPAgent behavior and LLM integration

Examples

Explore real-world configuration examples
New to mcp_use? Start with the Quickstart Guide for a basic introduction, then return here for detailed configuration options.