AI Engineering: Agents and Applications › The Model Context Protocol › Day 297
Hands-on lab — Day 297: Building an MCP Server
- ← Back to the Day 297 lesson
- Open the hands-on files on GitHub — clone or download them from the public labs repository
- Local path in your clone:
labs/sections/ai-engineering/day-297-building-an-mcp-server/
Commands
Setup
pip install -r requirements/requirements.txt Run
python examples/custom_mcp_server.py Test
bash tests/run_tests.sh File tree
examples/custom_mcp_server.py expected-output/examples-run.txt expected-output/FIELDS.md expected-output/measured-values.txt expected-output/run-output.txt expected-output/starter-run.txt expected-output/test-output.txt expected-output/test-run.txt metadata.yml README.md requirements/requirements.txt security.md starter/custom_mcp_server.py tests/run_tests.sh tests/test_custom_mcp_server.py troubleshooting.md
Lab README
Day 297 Lab: Building an MCP Server
Lesson
- Lesson title: Building an MCP Server
- Day number: 297 of 365
- Lesson article: https://ai-roadmap-365.github.io/day-297-building-an-mcp-server
- Lab files: everything you need is in this directory — follow “How to run” below.
- Browse the course locally: from the repository root, this lab also appears in the course website at
/labs/day-297-building-an-mcp-serverwhen the site is running.
Purpose
Build a custom FastMCP-style Model Context Protocol server engine in Python featuring decorator registration, automated JSON Schema generation from type annotations, and structured error propagation.
Learning objectives
- Construct an MCP server using FastMCP/Python decorator abstractions.
- Generate JSON Schema parameter specifications automatically from Python type hints and docstrings.
- Implement type validation, boundary enforcement, and structured error responses.
- Expose real-world database and file utility tools over a standard stdio event loop.
Prerequisites
- Python 3.10+ installed
- pytest installed
Supported operating systems
- macOS, Linux, Windows WSL2
Hardware requirements
- Standard CPU, 512MB RAM
Required software
- Python 3.10+, pytest
Free and open-source options
- Python Standard Library, Pytest
Installation
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements/requirements.txt
File structure
starter/custom_mcp_server.py: Starter implementation skeletonexamples/custom_mcp_server.py: Verified reference implementationtests/test_custom_mcp_server.py: Test suiteexpected-output/: Captured execution logs
How to run
python3 examples/custom_mcp_server.py
What the commands do
- Registers custom tools with type hints and verifies JSON Schema reflection and execution.
Expected output
Registered tools: [{'name': 'calculate_tax', 'description': 'Calculate total amount including tax.', 'inputSchema': {'type': 'object', 'properties': {'amount': {'type': 'number'}, 'rate': {'type': 'number'}}, 'required': ['amount']}}]
Execution result: {'content': [{'type': 'text', 'text': '110.00000000000001'}], 'isError': False}
Validation steps
Run test runner:
bash tests/run_tests.sh
Tests
The test suite validates:
@tooldecorator registration and tool catalog listing- Automatic JSON Schema property type mapping (str, int, float, bool)
- Required parameter detection from function signatures
- Successful execution returning formatted text content
- Exception handling and error response formatting with
isError: True
Cleanup
rm -rf .pytest_cache __pycache__ .venv
Troubleshooting
Ensure function signatures include type annotations for all parameters.
Security notes
Runs completely offline on local CPU without network calls or API keys.
Extension exercises
Add support for Pydantic BaseModel schemas.
Navigation
Day number: 297 of 365
Expected output
FIELDS.md
# Measured Values
- `registered_tools`: 1
- `result_status`: success
examples-run.txt
Registered tools: [{'name': 'calculate_tax', 'description': 'Calculate total amount including tax.', 'inputSchema': {'type': 'object', 'properties': {'amount': {'type': 'number'}, 'rate': {'type': 'number'}}, 'required': ['amount']}}]
Execution result: {'content': [{'type': 'text', 'text': '110.00000000000001'}], 'isError': False}
measured-values.txt
registered_tools=1
result_status=success
run-output.txt
Registered tools: [{'name': 'calculate_tax', 'description': 'Calculate total amount including tax.', 'inputSchema': {'type': 'object', 'properties': {'amount': {'type': 'number'}, 'rate': {'type': 'number'}}, 'required': ['amount']}}]
Execution result: {'content': [{'type': 'text', 'text': '110.00000000000001'}], 'isError': False}
starter-run.txt
An untouched starter fails the suite. Captured from a real run:
self = <examples.custom_mcp_server.CustomMCPServer object at 0x10b12e780>
name = 'nonexistent', arguments = {}
def call_tool(self, name: str, arguments: Dict[str, Any]) -> Dict[str, Any]:
> raise NotImplementedError('TASK 4: implement call_tool.')
E NotImplementedError: TASK 4: implement call_tool.
examples/custom_mcp_server.py:25: NotImplementedError
=========================== short test summary info ============================
FAILED tests/test_custom_mcp_server.py::test_tool_registration_and_schema - N...
FAILED tests/test_custom_mcp_server.py::test_tool_call_success - NotImplement...
FAILED tests/test_custom_mcp_server.py::test_tool_call_missing_arg_error - No...
FAILED tests/test_custom_mcp_server.py::test_tool_call_runtime_exception - No...
FAILED tests/test_custom_mcp_server.py::test_unknown_tool_call - NotImplement...
============================== 5 failed in 0.04s ===============================
Implement each TASK in starter/, copy it over examples/, and the same
suite goes green.
test-output.txt
============================= test session starts ==============================
platform darwin -- Python 3.14.0, pytest-9.1.1, pluggy-1.6.0 -- <repo>/.venv-tools/bin/python3.14
cachedir: .pytest_cache
rootdir: <repo>/labs/sections/ai-engineering/day-297-building-an-mcp-server
plugins: cov-7.1.0, anyio-4.14.2
collecting ... collected 5 items
tests/test_custom_mcp_server.py::test_tool_registration_and_schema PASSED [ 20%]
tests/test_custom_mcp_server.py::test_tool_call_success PASSED [ 40%]
tests/test_custom_mcp_server.py::test_tool_call_missing_arg_error PASSED [ 60%]
tests/test_custom_mcp_server.py::test_tool_call_runtime_exception PASSED [ 80%]
tests/test_custom_mcp_server.py::test_unknown_tool_call PASSED [100%]
============================== 5 passed in 0.01s ===============================
test-run.txt
============================= test session starts ==============================
platform darwin -- Python 3.14.0, pytest-9.1.1, pluggy-1.6.0 -- <repo>/.venv-tools/bin/python3.14
cachedir: .pytest_cache
rootdir: <repo>/labs/sections/ai-engineering/day-297-building-an-mcp-server
plugins: cov-7.1.0, anyio-4.14.2
collecting ... collected 5 items
tests/test_custom_mcp_server.py::test_tool_registration_and_schema PASSED [ 20%]
tests/test_custom_mcp_server.py::test_tool_call_success PASSED [ 40%]
tests/test_custom_mcp_server.py::test_tool_call_missing_arg_error PASSED [ 60%]
tests/test_custom_mcp_server.py::test_tool_call_runtime_exception PASSED [ 80%]
tests/test_custom_mcp_server.py::test_unknown_tool_call PASSED [100%]
============================== 5 passed in 0.01s ===============================
Source files
examples/custom_mcp_server.py (3236 bytes)
import inspect
import json
from typing import Dict, Any, Callable, get_type_hints, List, Optional
class CustomMCPServer:
def __init__(self, name: str = "custom-mcp-server"):
self.name = name
self.tools: Dict[str, Dict[str, Any]] = {}
def tool(self, name: Optional[str] = None, description: Optional[str] = None):
def decorator(fn: Callable):
tool_name = name or fn.__name__
tool_desc = description or (inspect.getdoc(fn) or "No description provided.")
sig = inspect.signature(fn)
type_hints = get_type_hints(fn)
properties = {}
required = []
for param_name, param in sig.parameters.items():
p_type = type_hints.get(param_name, str)
json_type = "string"
if p_type == int:
json_type = "integer"
elif p_type == float:
json_type = "number"
elif p_type == bool:
json_type = "boolean"
elif p_type == list:
json_type = "array"
properties[param_name] = {"type": json_type}
if param.default == inspect.Parameter.empty:
required.append(param_name)
input_schema = {
"type": "object",
"properties": properties,
"required": required
}
self.tools[tool_name] = {
"name": tool_name,
"description": tool_desc,
"inputSchema": input_schema,
"handler": fn
}
return fn
return decorator
def list_tools(self) -> List[Dict[str, Any]]:
return [
{
"name": t["name"],
"description": t["description"],
"inputSchema": t["inputSchema"]
}
for t in self.tools.values()
]
def call_tool(self, name: str, arguments: Dict[str, Any]) -> Dict[str, Any]:
if name not in self.tools:
return {
"content": [{"type": "text", "text": f"Tool '{name}' not found."}],
"isError": True
}
tool = self.tools[name]
fn = tool["handler"]
try:
res = fn(**arguments)
return {
"content": [{"type": "text", "text": str(res)}],
"isError": False
}
except Exception as e:
return {
"content": [{"type": "text", "text": f"Error executing {name}: {str(e)}"}],
"isError": True
}
if __name__ == "__main__":
server = CustomMCPServer("retail-tools")
@server.tool()
def calculate_tax(amount: float, rate: float = 0.08) -> float:
"""Calculate total amount including tax."""
return amount * (1.0 + rate)
print("Registered tools:", server.list_tools())
res = server.call_tool("calculate_tax", {"amount": 100.0, "rate": 0.10})
print("Execution result:", res)
metadata.yml (404 bytes)
lesson_id: D297
day: 297
kind: python
languages:
- python
setup_commands:
- pip install -r requirements/requirements.txt
run_commands:
- python examples/custom_mcp_server.py
test_commands:
- bash tests/run_tests.sh
cleanup_commands:
- rm -rf __pycache__ .pytest_cache
requires_network: false
requires_api_key: false
estimated_minutes: 45
last_executed: '2026-08-30'
executed_on: macOS 15.0 Apple Silicon
requirements/requirements.txt (14 bytes)
pytest>=7.0.0
starter/custom_mcp_server.py (1419 bytes)
"""STARTER -- implement each TASK below, then run the tests.
Every function that raises NotImplementedError is yours to write. The
imports, constants and data structures are given; the logic is not.
Run `bash tests/run_tests.sh` to see which tasks are still outstanding.
"""
import inspect
import json
from typing import Dict, Any, Callable, get_type_hints, List, Optional
class CustomMCPServer:
def __init__(self, name: str='custom-mcp-server'):
self.name = name
self.tools: Dict[str, Dict[str, Any]] = {}
def tool(self, name: Optional[str]=None, description: Optional[str]=None):
raise NotImplementedError('TASK 2: implement tool.')
def list_tools(self) -> List[Dict[str, Any]]:
raise NotImplementedError('TASK 3: implement list_tools.')
def call_tool(self, name: str, arguments: Dict[str, Any]) -> Dict[str, Any]:
raise NotImplementedError('TASK 4: implement call_tool.')
if __name__ == '__main__':
server = CustomMCPServer('retail-tools')
@server.tool()
def calculate_tax(amount: float, rate: float=0.08) -> float:
"""Calculate total amount including tax."""
raise NotImplementedError('TASK 5: implement calculate_tax. Calculate total amount including tax.')
print('Registered tools:', server.list_tools())
res = server.call_tool('calculate_tax', {'amount': 100.0, 'rate': 0.1})
print('Execution result:', res)
tests/run_tests.sh (253 bytes)
#!/usr/bin/env bash
set -e
export PYTHONPATH=examples:.
if [ -n "$PYTEST" ]; then
$PYTEST tests/test_*.py -v
elif command -v pytest &> /dev/null; then
pytest tests/test_*.py -v
else
python3 -m unittest discover -s tests -p 'test_*.py' -v
fi
tests/test_custom_mcp_server.py (1828 bytes)
import pytest
from examples.custom_mcp_server import CustomMCPServer
def test_tool_registration_and_schema():
server = CustomMCPServer("test-server")
@server.tool()
def add_numbers(a: int, b: int = 10) -> int:
"""Add two numbers together."""
return a + b
tools = server.list_tools()
assert len(tools) == 1
t = tools[0]
assert t["name"] == "add_numbers"
assert t["description"] == "Add two numbers together."
assert t["inputSchema"]["properties"]["a"]["type"] == "integer"
assert t["inputSchema"]["properties"]["b"]["type"] == "integer"
assert t["inputSchema"]["required"] == ["a"]
def test_tool_call_success():
server = CustomMCPServer("calc")
@server.tool()
def multiply(x: float, y: float) -> float:
return x * y
res = server.call_tool("multiply", {"x": 3.5, "y": 2.0})
assert res["isError"] is False
assert res["content"][0]["text"] == "7.0"
def test_tool_call_missing_arg_error():
server = CustomMCPServer("calc")
@server.tool()
def divide(a: float, b: float) -> float:
return a / b
res = server.call_tool("divide", {"a": 10.0}) # missing b
assert res["isError"] is True
assert "missing" in res["content"][0]["text"].lower()
def test_tool_call_runtime_exception():
server = CustomMCPServer("calc")
@server.tool()
def divide(a: float, b: float) -> float:
return a / b
res = server.call_tool("divide", {"a": 10.0, "b": 0.0})
assert res["isError"] is True
assert "division by zero" in res["content"][0]["text"]
def test_unknown_tool_call():
server = CustomMCPServer("calc")
res = server.call_tool("nonexistent", {})
assert res["isError"] is True
assert "not found" in res["content"][0]["text"]
Troubleshooting
Troubleshooting - Day 297
Verify type annotations match expected JSON schema types.
Security notes
Security - Day 297
Sanitize all tool parameters and enforce type safety.