cuda-pathfinder

Pathfinder for CUDA components

cuda-pathfinder Permissive license Apache-2.0 Active ★ 3,341 v1.6.0 released 2026-07-21

Install

cuda-pathfinder on PyPI

pip

pip install cuda-pathfinder

uv

uv add cuda-pathfinder

poetry

poetry add cuda-pathfinder

Package facts

License	Apache-2.0 (permissive)
Python support	supports the current Python release (>=3.10)
Install friction	low — pure-Python wheel
Runtime dependencies	none
Maintenance	actively maintained — 23 days since the last release
Last repo commit	2026-08-13
First released	2025-07-16
Popularity	one of the top 1,000 most-downloaded packages on PyPI (30-day window, as of 2026-08-13)
Known vulnerabilities	none known (OSV.dev, checked 2026-08-13)

Evidence: cuda_pathfinder-1.6.0-py3-none-any.whl

About cuda-pathfinder

from the package's own PyPI description — quoted content, verbatim

cuda-pathfinder: Utilities for locating CUDA components

.. image:: https://img.shields.io/badge/NVIDIA-black?logo=nvidia :target: https://www.nvidia.com/ :alt: NVIDIA

cuda.pathfinder <https://nvidia.github.io/cuda-python/cuda-pathfinder/>_ aims to be a one-stop solution for locating CUDA components. Currently it supports locating and loading dynamic libraries (.so, .dll), and locating CTK header directories. Support for other artifacts is in progress.

Documentation <https://nvidia.github.io/cuda-python/cuda-pathfinder/>_
Releases <https://nvidia.github.io/cuda-python/cuda-pathfinder/latest/release.html>_
Repository <https://github.com/NVIDIA/cuda-python/tree/main/cuda_pathfinder/>_
Issue tracker <https://github.com/NVIDIA/cuda-python/issues/>_ (select component cuda.pathfinder)

cuda.pathfinder is under active development. Feedback and suggestions are...

Read as markdown · JSON record · Source repository · Docs

AI interpretation — verify before relying

AI-generated interpretation of the package facts above; every digit, version, license, or vulnerability id it cites is grounded in the facts already shown on this page

Locates and loads CUDA Toolkit components—dynamic libraries (.so, .dll) and CTK header directories—providing a unified interface for discovering CUDA artifacts across different system configurations.

Installs with no runtime dependencies and low friction. Actively maintained by NVIDIA with recent commits and a top-1000 PyPI tier presence; Python 3.10+ required.

Licensed under Apache-2.0 (permissive), allowing commercial and private use with minimal restrictions; attribution required.

Usage

pip install cuda-pathfinder

import cuda.pathfinder
# Locate CUDA libraries and headers
libs = cuda.pathfinder.find_libraries()
headers = cuda.pathfinder.find_headers()

Requires Python 3.10 or later; CUDA Toolkit installation on the system is assumed.

Verdict: A lightweight, actively maintained utility from NVIDIA for discovering CUDA components across platforms. Zero runtime dependencies, permissive licensing, and no known vulnerabilities make it a low-risk choice for projects needing programmatic CUDA artifact discovery.

Needs verification

Whether the package requires a specific CUDA Toolkit version or works with all supported CTK versions as claimed
Actual API surface and whether the example calls (find_libraries, find_headers) are the correct public interface

cuda toolkit component locatorfind cuda libraries dynamicallycuda header directory discoverynvidia cuda pathfinding utilitylocate cuda shared objectscuda ctk path resolver