PySTAC

PySTAC is a library for working with the SpatioTemporal Asset Catalog specification in Python 3.

Installation

Install from PyPi (recommended)

python -m pip install pystac

If you would like to enable the validation feature utilizing the jsonschema project, install with the optional validation requirements:

python -m pip install 'pystac[validation]'

If you would like to use the orjson instead of the standard json library for JSON serialization/deserialization, install with the optional orjson requirements:

python -m pip install 'pystac[orjson]'

If you would like to use a custom RetryStacIO class for automatically retrying network requests when reading with PySTAC, or if you have non-ASCII characters in your urls you'll need urllib3:

python -m pip install 'pystac[urllib3]'

If you are using jupyter notebooks and want to enable pretty display of pystac objects you'll need jinja2

python -m pip install 'pystac[jinja2]'

Install from source

git clone https://github.com/stac-utils/pystac.git
cd pystac
python -m pip install .

See the installation page for more options.

Versioning

The core PySTAC API follows Semantic Versioning. As of Spring 2026, our extension implementations have moved to their own Python packages, so they can be versioned independently of the core PySTAC API. Breaking changes to the API of extension packages are not considered breaking changes for the purposes of PySTAC versioning.

The STAC extension packages' version will match the version extension itself, e.g. pystac-ext-projection v2.0.0 corresponds to the v2.0.0 release of the projection extension. Any changes to the software (not the extension) will be released via post releases.

Documentation

See the documentation page for the latest docs.

Developing

See contributing docs for details on contributing to this project.

Running the quickstart and tutorials

There is a quickstart and tutorials written as jupyter notebooks in the docs/tutorials folder. To run the notebooks, run a jupyter notebook with the docs directory as the notebook directory:

jupyter notebook --ip 0.0.0.0 --port 8888 --notebook-dir=docs

You can then navigate to the notebooks and execute them.

Requires Jupyter be installed.