Skip to content

Guide · Agent skill

How to use the UV Package Manager Guide agent skill

Updated · written with AI from the project's own documentation and checked against our catalogue data

The uv-package-manager skill gives your AI agent a written guide to uv, a Python package installer and environment manager written in Rust. Once it is loaded, the agent can give you step-by-step commands for creating projects, adding and removing libraries, locking dependencies and installing Python versions. This guide explains what the skill does, how to install it, how to ask for help with natural requests, and what to watch for.

UV Package Manager Guidewshobson/agents · MIT

Quick start

  1. 1Install uv. Install uv with Homebrew, Cargo, WinGet or the official installer, then confirm it with uv --version.
  2. 2Download the skill. Download the uv-package-manager ZIP from this page and unzip it.
  3. 3Place the folder. Move it to ~/.claude/skills/ for personal use or to .claude/skills/ in a project, or upload it under Settings → Capabilities → Skills in the Claude apps.
  4. 4Ask in plain language. Describe your Python task, such as setting up a project or migrating from requirements.txt, and mention uv.
  5. 5Review commands. Read each suggested command before running it, especially ones that add dependencies or change project files.
  6. 6Lock and sync. Use uv lock to create uv.lock and uv sync to install from it so installs stay consistent.

What is an agent skill, and what does this one do?

An agent skill is a folder of instructions that an AI agent loads when a task matches its description. This skill is a written guide to the uv tool. It does not add new abilities to your computer; it gives the agent reference patterns and commands to draw on.

With the skill loaded, the agent can help you create new Python projects, manage dependencies faster than pip, create and manage virtual environments, install Python interpreters, work with lockfiles for reproducible builds, migrate from pip, pip-tools or poetry, and optimize Docker and CI builds.

Install uv before you install the skill

The skill does not install uv on its own, so uv must already be on your machine. The source lists these options: Homebrew on macOS or Linux, Cargo if you have Rust, WinGet on Windows, or the official installer described in the uv installation guide.

To check the installation, run the version command below in a terminal.

  • Homebrew (macOS or Linux): brew install uv
  • Cargo (if you have Rust): cargo install --locked uv
  • Windows: use WinGet, or the official installation guide for other methods
uv --version

How do you install the uv-package-manager skill?

Download the ZIP file for this skill from our page and unzip it. For a personal setup, place the unzipped folder in ~/.claude/skills/. For a single project, place it in that project's .claude/skills/ folder.

If you use the Claude apps, you can instead upload the skill under Settings → Capabilities → Skills.

  • Personal: ~/.claude/skills/uv-package-manager/
  • Project only: .claude/skills/uv-package-manager/ inside your project
  • Claude apps: Settings → Capabilities → Skills, then upload

How do you trigger the skill with natural requests?

You do not need a special command. Describe the Python task you want done, and the agent will use the skill when your request matches its description, which covers setting up Python projects, managing dependencies and optimizing Python workflows with uv.

Name uv in your request when you want the guide used for certain. Include your existing files when you are migrating, so the agent can work from them.

  • Set up a new Python project with uv and Python 3.12
  • Migrate my requirements.txt to uv and explain each command
  • Add pytest and ruff as dev dependencies using uv
  • Write a Dockerfile for this Python app that uses uv for fast builds
  • Show me how to lock dependencies so CI installs the same versions

Commands the skill covers

The guide groups its patterns by task. Here are some of the core commands it shows. Check the official uv documentation before running them, because uv changes quickly.

Creating a project uses uv init, adding packages uses uv add, and running code without manual activation uses uv run.

  • uv init my-project creates a project with .python-version, pyproject.toml, README.md and .gitignore
  • uv add requests pandas adds packages; uv add --dev pytest adds dev dependencies
  • uv sync installs dependencies from pyproject.toml
  • uv lock generates the uv.lock file
  • uv run python app.py runs a script in the project environment
  • uv python install 3.12 installs a Python version
uv init my-project cd my-project uv add requests pandas uv add --dev pytest black ruff uv run python app.py

Advanced topics in the reference file

The main SKILL.md file covers the everyday workflow. The references/advanced-patterns.md file adds monorepo support with workspaces, CI/CD integration with GitHub Actions, Docker integration, troubleshooting, and a command reference.

Ask the agent to open that reference when your question goes beyond basic project setup.

  • Monorepo workspaces with [tool.uv.workspace]
  • GitHub Actions example using astral-sh/setup-uv
  • Docker, troubleshooting and command reference

Safety notes: scripts and network access

This skill runs no scripts and uses no network access. It is a written guide that the agent reads. The commands it suggests, such as uv add with a git URL or uv sync, will still act on your project files and may download packages when you run them yourself.

Review any command before running it, especially ones that add packages from git URLs or local paths, and keep your project under version control so you can see what changed.

  • No scripts are bundled to run
  • No network access by the skill itself
  • Commands you approve may download packages or change pyproject.toml and uv.lock

What people use it for

  • Starting a new Python project with a pyproject.toml, .python-version and dev tools
  • Migrating a requirements.txt or poetry project to uv
  • Adding a lockfile so CI and teammates get the same dependency versions
  • Writing a Docker or GitHub Actions setup that uses uv for faster installs
  • Installing and pinning a specific Python version for a project

Frequently asked questions

How do I install the uv package manager?

The source lists Homebrew (brew install uv) on macOS or Linux, Cargo (cargo install --locked uv) if you have Rust, and WinGet (winget install --id=astral-sh.uv -e) on Windows. For other methods, use the official installation guide linked from the skill.

Is the uv-package-manager skill free?

The skill is released under the MIT licence, as listed on this page. uv itself is a separate tool that you install on your machine.

Do I need uv installed before using this skill?

Yes. The skill does not install uv. It requires uv on the machine, and Python project files such as requirements.txt or pyproject.toml when you are migrating.

Does the skill run scripts or use the network?

No. The skill runs no scripts and uses no network access. Commands you run from its guidance may still download packages, so review them first.

How is uv different from pip or poetry?

The skill describes uv as faster than pip, compatible with pip workflows, and lighter than poetry. It also says uv works with pip-tools and poetry projects. The 10-100x speed claim is a general figure from the source, not a verified benchmark, so test it on your own project.

Why might the commands in the guide not work for me?

uv changes quickly and the guide shows a placeholder version, so syntax may differ from your installed release. Check uv --version and the official uv documentation, and ask the agent to confirm commands against current docs.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

This skill suits Python developers starting a new project, teams moving from pip, pip-tools or poetry to uv, and anyone who needs reproducible lockfiles or faster CI and Docker builds. It is not useful for non-Python projects, and if you need guaranteed-current command syntax, check the official uv documentation rather than relying on this guide alone.

Download the skill