Dummy MCP Server

A demonstration MCP (Model Context Protocol) server showcasing best practices for tool implementation, input validation, response formatting, pagination, and error handling.

Overview

This is a dummy MCP server built with Python and FastMCP that demonstrates:

• Best Practice Tool Design: Following MCP guidelines for naming, annotations, and structure
• Input Validation: Using Pydantic v2 models with comprehensive field constraints
• Multiple Response Formats: Supporting both JSON (machine-readable) and Markdown (human-readable) outputs
• Pagination: Implementing offset-based pagination with clear metadata
• Character Limits: Respecting character limits with graceful truncation
• Error Handling: Providing clear, actionable error messages
• Type Safety: Full type hints throughout the codebase

Available Tools

1. dummy_get_time

Get the current date and time.

Parameters:

• timezone (optional): Timezone for display (default: "UTC")
• response_format (optional): Output format ("markdown" or "json")

Example:

{
  "timezone": "UTC",
  "response_format": "json"
}

Annotations: Read-only, Idempotent

---

2. dummy_calculate

Perform basic mathematical operations (add, subtract, multiply, divide).

Parameters:

• operation: Mathematical operation ("add", "subtract", "multiply", "divide")
• a: First number
• b: Second number
• response_format (optional): Output format ("markdown" or "json")

Example:

{
  "operation": "add",
  "a": 5.0,
  "b": 3.0,
  "response_format": "json"
}

Annotations: Read-only, Idempotent

---

3. dummy_search_items

Search for dummy items with pagination and filtering.

Parameters:

• query (optional): Search query to filter items
• limit (optional): Maximum results to return (1-100, default: 20)
• offset (optional): Number of results to skip for pagination (default: 0)
• response_format (optional): Output format ("markdown" or "json")

Example:

{
  "query": "development",
  "limit": 10,
  "offset": 0,
  "response_format": "json"
}

Annotations: Read-only, Idempotent

---

4. dummy_create_item

Create a new dummy item.

Parameters:

• name: Name of the item (1-100 characters, required)
• description (optional): Description of the item (max 500 characters)
• tags (optional): List of tags (max 10 tags)
• response_format (optional): Output format ("markdown" or "json")

Example:

{
  "name": "Project Alpha",
  "description": "New Q1 project",
  "tags": ["development", "priority-1"],
  "response_format": "json"
}

Annotations: Non-destructive write operation

---

5. dummy_format_text

Format text using various transformation options.

Parameters:

• text: Text to format (1-10000 characters, required)
• format_type: Formatting to apply ("uppercase", "lowercase", "title", "reverse")

Example:

{
  "text": "hello world",
  "format_type": "uppercase"
}

Annotations: Read-only, Idempotent

---

Installation

Prerequisites

1. Install Python 3.10 or higher

2. Install uv (fast Python package installer):

# macOS/Linux
curl -LsSf https://astral.sh/uv/install.sh | sh

# Windows
powershell -c "irm https://astral.sh/uv/install.ps1 | iex"

# Or with pip
pip install uv

Install Dependencies

# Install dependencies using uv
uv sync

# Or install manually
uv pip install -e .

Usage

Running the Server

The server uses stdio transport by default, which is suitable for command-line tools and subprocess integration:

# Run with uv (recommended - manages dependencies automatically)
uv run server.py

# Or run directly if dependencies are installed
python3 server.py

Testing with MCP Inspector

You can test the server using the MCP Inspector tool:

# With uv (recommended)
npx @modelcontextprotocol/inspector uv run server.py

# Or without uv
npx @modelcontextprotocol/inspector python3 server.py

Integration with Claude Desktop

Add this configuration to your Claude Desktop config file:

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

Using uv (recommended):

{
  "mcpServers": {
    "dummy": {
      "command": "uv",
      "args": [
        "--directory",
        "/path/to/dummy-mcp",
        "run",
        "server.py"
      ]
    }
  }
}

Or without uv:

{
  "mcpServers": {
    "dummy": {
      "command": "python3",
      "args": ["/path/to/dummy-mcp/server.py"]
    }
  }
}

Architecture

Server Structure

dummy-mcp/
├── server.py          # Main MCP server implementation
├── pyproject.toml     # Project metadata and dependencies (uv)
├── requirements.txt   # Python dependencies (legacy, optional)
└── README.md         # This file

Key Components

Pydantic Models: All tool inputs are validated using Pydantic v2 models with:

• Field constraints (min/max length, ranges)
• Custom validators
• Type safety
• Automatic error messages

Shared Utilities: Common functionality is extracted into reusable functions:

• _generate_dummy_items(): Generate test data
• _filter_items(): Filter items by query
• _format_items_markdown(): Format as Markdown
• _format_items_json(): Format as JSON
• _check_character_limit(): Enforce character limits

Tool Annotations: All tools include proper annotations:

• readOnlyHint: Indicates if tool modifies state
• destructiveHint: Indicates if modifications are destructive
• idempotentHint: Indicates if repeated calls have same effect
• openWorldHint: Indicates if tool interacts with external systems

Best Practices Demonstrated

This server follows MCP best practices:

1. Server Naming: dummy_mcp follows Python convention {service}_mcp
2. Tool Naming: Snake_case with service prefix (e.g., dummy_get_time)
3. Response Formats: Supports both JSON and Markdown
4. Pagination: Implements offset-based pagination with metadata
5. Character Limits: 25,000 character limit with graceful truncation
6. Input Validation: Comprehensive Pydantic models
7. Error Handling: Clear, actionable error messages
8. Documentation: Comprehensive docstrings with examples
9. Type Safety: Full type hints throughout
10. Code Reusability: Shared utilities for common operations

Development

Verifying Syntax

Check Python syntax:

python3 -m py_compile server.py

Running Tests

# Install dev dependencies
uv sync --all-extras

# Run tests (when available)
uv run pytest

Code Quality

The implementation includes:

• Type hints throughout
• Comprehensive docstrings
• Pydantic validation
• Error handling
• Shared utilities for DRY code

License

This is a demonstration project for educational purposes.

Resources

• MCP Documentation
• MCP Python SDK
• FastMCP Documentation
• uv Documentation