# Model Context Protocol (MCP)

# example

# Cline

version: 3.11.0

open vscode and install cline extension
open Cline setting, API Provider choose OpenRouer
Model choose deepseek/deepseek-chat:free
click Get OpenRouter API Key（if you have）

mcp_set_openroter

test it

mcp_test_openrouter

click MCP Servers find time and install
open new terminal pip install mcp-server-time
it will create cline_mcp_settings.json

# cline_mcp_settings.json

i used conda, so maybe the command string is different


{
  "mcpServers": {
    "github.com/modelcontextprotocol/servers/tree/main/src/time": {
      "command": "/opt/anaconda3/envs/mcp_time/bin/python",
      "args": ["-m", "mcp_server_time", "--local-timezone=Asia/Shanghai"],
      "disabled": false,
      "autoApprove": []
    }
  }
}

# what kind of protocol?

we need ai-gateway of cloudflare

dash > AI > AI Gateway > create new one
click this that it is your created and find API button
Platform select OpenRouer & copy API Endpoint url
go back your Cline setting and fall it

mcp_set_cloudflare

new task then ask 'what time now'

open AI Gateway logs

mcp_gatway

copy Request logs, you will find it in system prompt


You have the ability to create an MCP server and add it to a configuration file that will then expose the tools and resources for you to use with `use_mcp_tool` and `access_mcp_resource`). The documentation provides detailed information about the MCP server creation process, including setup instructions, best practices, and examples.

copy Response logs, OpenAI give your answer is use the use_mcp_tool tood


"<thinking>\nThe user wants to know the current time. I can use the `use_mcp_tool` tool to get the current time in the Asia/Shanghai timezone.\n</thinking>\n\n<use_mcp_tool>\n<server_name>github.com/modelcontextprotocol/servers/tree/main/src/time</server_name>\n<tool_name>get_current_time</tool_name>\n<arguments>\n{\n  \"timezone\": \"Asia/Shanghai\"\n}\n</arguments>\n</use_mcp_tool>"

# process

mcp_run_process

# Function Calling vs MCP

Feature	Function Calling	MCP
Purpose	Converts user prompts into structured function calls.	Standardizes execution and response handling.
Who Controls It?	LLM provider	External system handling LLM integration.
Output Format	Varies by LLM vendor (JSON-based).	Uses a standardized protocol (e.g., JSON-RPC).
Flexibility	Different models structure calls differently.	Ensures interoperability across multiple tools.

# other

origin news (opens new window)

introduction mcp (opens new window)

mcp VS function calling (opens new window)

https://smithery.ai/ (opens new window)

time mcp ofgithub (opens new window)

https://github.com/astral-sh/uv/ (opens new window)

if you had Apple could not verify “uvx” is free of malware that may harm your Mac or compromise your privacy. open Mac Setting pravicy & security click allow anyway

ai-gatway document (opens new window)

← AI local