mcp-dev-utils

A lightweight Model Context Protocol server that exposes everyday developer utilities as MCP tools and resources.

Built with mcp — the official Python MCP SDK (uses FastMCP for a clean, decorator-based API).

---

Tools

Tool	Description
format_json	Pretty-print and validate a JSON string
generate_uuid	Generate 1–20 random UUID v4 values
base64_convert	Encode or decode Base64 strings
http_status	Look up an HTTP status code name and description
unix_timestamp	Convert a Unix timestamp to UTC or get the current time

Resources

URI	Description
utils://http-status-codes	Full reference list of supported HTTP status codes

---

Running locally

Requirements: Python ≥ 3.11, uv (or pip)

# Install dependencies
uv sync

# Run over stdio (default — used by most MCP hosts)
uv run python server.py

# Or run as an SSE server on port 8000
uv run python -c "from server import mcp; mcp.run(transport='sse')"

# Test interactively in the browser
uv run mcp dev server.py

Connecting to Claude Desktop

Add the following to your claude_desktop_config.json:

{
  "mcpServers": {
    "dev-utils": {
      "command": "uv",
      "args": ["run", "python", "server.py"],
      "cwd": "/absolute/path/to/mcp-dev-utils"
    }
  }
}

On macOS the config lives at: ~/Library/Application Support/Claude/claude_desktop_config.json

On Windows the config lives at: %APPDATA%\Claude\claude_desktop_config.json

---

Project structure

mcp-dev-utils/
├── server.py        # All tools, resources, and server definition
├── test_client.py   # Example MCP client for manual testing
├── test_msg.txt     # Raw JSON-RPC messages (educational — use test_client.py for reliable testing)
├── pyproject.toml   # Project metadata and dependencies
└── README.md

Key design decisions

• Single-file server — keeps the codebase easy to read and audit at a glance; ideal for a portfolio piece where clarity matters.
• FastMCP — the high-level SDK wrapper that removes boilerplate while still showing the underlying MCP concepts (tools, resources, structured arguments).
• stdio transport — the default and most universally supported transport; no extra infrastructure needed.
• No external dependencies — all tools rely only on the Python standard library plus mcp itself.