Pixi Cross-Platform Package Manager Built on Conda
A blazing-fast, cross-platform package manager and workflow tool written in Rust. Pixi builds on the Conda ecosystem to provide reproducible, multi-language dependency management with a Cargo-like developer experience for Python, C++, R, and more.
A blazing-fast, cross-platform package manager and workflow tool written in Rust. Pixi builds on the Conda ecosystem to provide reproducible, multi-language dependency management with a Cargo-like developer experience for Python, C++, R, and more.
npx skills add agentskillexchange/skills --skill pixi-cross-platform-package-manager-conda-rust
Pixi is a modern package manager from prefix.dev, written entirely in Rust, that brings the reproducibility of the Conda ecosystem together with the ergonomics of tools like Cargo and npm. It manages project dependencies, virtual environments, and task execution across Python, C++, R, CUDA, and other languages through a single pixi.toml configuration file.
The core commands mirror familiar workflows: pixi init creates a new project, pixi add numpy adds a dependency (resolving it from conda-forge or PyPI), pixi install locks and installs all dependencies, and pixi run executes tasks in the project environment. Every project gets an automatically maintained lock file (pixi.lock) that pins exact package versions and hashes for full reproducibility.
Key differentiators include: multi-language support — unlike pip/npm which are single-language, Pixi resolves Python, C++, R, Fortran, CUDA, and system libraries from the conda-forge repository containing 30,000+ packages; cross-platform environments with native support for Linux, macOS (Intel and Apple Silicon), and Windows without containers; PyPI integration allowing mixed conda and pip dependencies in the same project; and global tool installation via pixi global install for system-wide CLI tools.
The task system supports defining named commands in pixi.toml with dependency ordering, platform-specific overrides, and environment selection. Multiple named environments can coexist in a single project (e.g., separate test, docs, and dev environments with different dependency sets).
Built on the rattler library (also from prefix.dev), Pixi achieves fast dependency resolution through a SAT solver and parallel downloads. It supports conda channels, PyPI indices, and custom package repositories. The project has 6,500+ GitHub stars, 450+ forks, a BSD-3-Clause license, regular releases, and an active community.
Common use cases include data science projects with mixed Python/C++ dependencies, machine learning workflows requiring CUDA toolkit management, cross-platform CI/CD pipelines, and any project where “works on my machine” problems need to be eliminated through deterministic dependency resolution.