Introducing uv: A High‑Performance Unified Python Packaging Tool

uv is a high‑performance Python package manager written in Rust, designed as a drop‑in replacement for pip and as a comprehensive solution for managing Python projects, command‑line tools, single‑file scripts, and even the Python interpreter itself.

Key new capabilities include:

End‑to‑end project management: uv run , uv lock and uv sync create cross‑platform lock files and replace tools like Poetry, PDM, and Rye.

Tool management: uv tool install (alias uvx ) installs CLI tools in isolated environments and can execute them without prior installation, offering a fast alternative to pipx.

Python installation: uv python install can fetch and install specific Python versions, removing the need for external managers such as pyenv.

Workspace support: Inspired by Cargo, uv can manage a collection of packages in a single workspace, sharing a common lock file while each package retains its own pyproject.toml .

Dependency sources: uv extends PEP 621 with tool.uv.sources to declare editable or path‑based dependencies, enabling local development without changing project metadata.

Single‑file script execution: Scripts can embed dependency metadata (PEP‑723) and be run directly with uv run or uvx , automatically creating a temporary virtual environment.

Installation is straightforward via the official installer:

<code>curl -LsSf https://astral.sh/uv/install.sh | sh</code>

On Windows, use PowerShell:

<code>powershell -c "irm https://astral.sh/uv/install.ps1 | iex"</code>

uv can also be installed from PyPI:

<code>pip install uv</code>

Example project initialization:

<code>uv init && uv add "fastapi>=0.112"</code>

This generates a pyproject.toml with the specified dependencies and allows uv lock to produce a lock file that works across macOS, Linux, and Windows.

For workspace configuration, a root pyproject.toml may contain:

<code>[project]
name = "fastapi"
version = "0.1.0"
dependencies = ["uvicorn"]

[tool.uv.sources]
uvicorn = { workspace = true }

[tool.uv.workspace]
members = ["libraries/*"]
</code>

Running a package within the workspace is as simple as:

<code>uv run --package fastapi</code>

uv’s tool API also provides commands to list and upgrade installed tools:

<code>uv tool list
uv tool upgrade --all</code>

Single‑file script example (main.py) with embedded dependencies:

<code># /// script
# requires-python = ">=3.12"
# dependencies = ["requests<3", "rich"]
# ///
import requests
from rich.pretty import pprint
resp = requests.get("https://peps.python.org/api/peps.json")
data = resp.json()
for k, v in list(data.items())[:10]:
    pprint((k, v["title"]))
</code>

Running the script with uv run main.py automatically resolves and installs the declared dependencies in an isolated environment.

Overall, uv offers a fast, reliable, and unified workflow for Python development, from simple scripts to large multi‑package workspaces, while remaining cross‑platform and self‑contained.