MCP Integration API

Tool integration via Model Context Protocol

This document covers AgenticGoKit's MCP (Model Context Protocol) integration API, which enables agents to discover, connect to, and use external tools and services. MCP provides a standardized way to integrate with various tools, from web search to database operations.

📋 Core Concepts

MCP Overview

MCP (Model Context Protocol) is a protocol for connecting AI agents with external tools and services. AgenticGoKit provides comprehensive MCP integration with three levels of complexity:

• Basic MCP: Simple tool discovery and execution
• Enhanced MCP: Caching and performance optimization
• Production MCP: Enterprise-grade features with monitoring and scaling

🚀 Basic Usage

CLI Integration

The easiest way to get started with MCP is using AgentCLI:

# Create project with MCP enabled
agentcli create my-project --enable-mcp

# Create with specific MCP level
agentcli create my-project --mcp minimal    # Basic MCP
agentcli create my-project --mcp standard   # With caching
agentcli create my-project --mcp advanced   # Full enterprise features

For complete CLI documentation, see the MCP CLI Guide.

Quick Start with MCP

// Initialize MCP with automatic discovery
err := core.QuickStartMCP()
if err != nil {
    panic(fmt.Sprintf("Failed to initialize MCP: %v", err))
}

// Create an MCP-aware agent
llmProvider, err := core.NewOpenAIProvider()
if err != nil {
    panic(fmt.Sprintf("Failed to create LLM provider: %v", err))
}

agent, err := core.NewMCPAgent("assistant", llmProvider)
if err != nil {
    panic(fmt.Sprintf("Failed to create MCP agent: %v", err))
}

📚 Configuration Reference

MCP Levels Comparison

Level	CLI Flag	Caching	Metrics	Load Balancing	Use Case
Minimal	--mcp minimal	❌	❌	❌	Development, Simple Apps
Standard	--mcp standard	✅	❌	❌	Production Applications
Advanced	--mcp advanced	✅	✅	✅	Enterprise Systems

Generated Configuration

When using AgentCLI with MCP flags, the following configuration is generated in agentflow.toml:

[mcp]
enabled = true
transport = "tcp"
enable_discovery = true
connection_timeout = 5000
max_retries = 3
retry_delay = 1000

# Additional features based on MCP level:
# --mcp standard adds:
enable_caching = true
cache_timeout = 300000

# --mcp advanced adds:
enable_metrics = true
enable_load_balancing = true
max_connections = 10

# Tool server examples (commented by default)
[[mcp.servers]]
name = "docker-http-sse"
type = "http_sse"
host = "localhost"
port = 8812
enabled = false

[[mcp.servers]]
name = "brave-search"
type = "stdio"
command = "npx @modelcontextprotocol/server-brave-search"
enabled = false

For complete documentation including server discovery, caching, production deployment, and custom tool development, see the Agent API reference.