[Python] When would you need to define your own custom MCP client?

I'm new to MCP. Particularly im looking to implement an agentic service using FastMCP in python. From what I understood from the docs the llm (whatever api/sdk/framework your using) is the client as most of the major ones support mcp (e.g. anthropic messages api). As in you do not need to do something like this:

import asyncio
from fastmcp import Client, FastMCP

# In-memory server (ideal for testing)
server = FastMCP("TestServer")
client = Client(server) # <- this is what im referring to

async def main():
async with client:
# Basic server interaction
await client.ping()

# List available operations
tools = await client.list_tools()
resources = await client.list_resources()
prompts = await client.list_prompts()

# Execute operations
result = await client.call_tool("example_tool", {"param": "value"})
print(result)

asyncio.run(main())


Instead you would do something like this

import anthropic
from rich import print

# Your server URL (replace with your actual URL)
url = 'https://your-server-url.com'

client = anthropic.Anthropic()

response = client.beta.messages.create(
model="claude-sonnet-4-20250514",
max_tokens=1000,
messages=[{"role": "user", "content": "Roll a few dice!"}],
mcp_servers=[
{
"type": "url",
"url": f"{url}/mcp/",
"name": "dice-server",
}
],
extra_headers={
"anthropic-beta": "mcp-client-2025-04-04"
}
)

print(response.content)


Where I interpret the above as that the anthropic messages API is the client, you don't need to explicitly define client = Client(server)

So I'm wondering what scenarios you would need to explicitly define your own mcp client when working with llms. (I can see the use of a client if you need to verify responses from servers but wondering about other cases). I may just be misunderstanding it entirely so would appreciate clarification

Continue reading...