Установка Python и uv

Before you build any MCP server, you need a tidy Python project. A clean start means fewer surprises and fewer broken installs later. Let's set the foundation. 🧱

The MCP Python SDK needs a modern interpreter, so you want Python 3.10 or newer. Older versions miss the typing features the SDK relies on.

Instead of juggling pip and virtualenv, you'll use uv, a single fast tool that manages Python, packages, and environments for you in one place.

uv replaces pip, pip-tools, pyenv, and virtualenv at once. Learning one command set keeps your MCP workflow simple and consistent across every project. 🚀

On macOS or Linux you can install uv with a single curl command. It drops a tiny binary on your path, ready to use immediately.

After installing, run a quick version check. If you see a version number, uv is on your path and ready to drive the rest of your setup.

You don't even need Python installed first. With uv python install you can fetch any version uv manages, so your project is never tied to the system Python.

Spin up a fresh project folder with uv init. It scaffolds the files you need so you can start adding MCP code right away.

That command writes a pyproject.toml to declare your project and its dependencies. This file becomes the single source of truth for what your server needs.

When you add a package, uv quietly creates a .venv folder, an isolated environment just for this project. Your global Python stays untouched and clean.

You never have to activate the venv by hand. uv run executes commands inside your project's environment automatically, so the right Python is always used.

You're setting up a fresh MCP project from zero.

You installed uv, let it manage Python, and scaffolded a clean project. With pyproject.toml and an isolated venv ready, you can now add the MCP SDK. ✅