Cerberus documents a test command that needs a setup.py the repository does not contain
Lightweight, extensible data validation library for Python
At a glance
- What is it?
- Cerberus is a small schema validation library for Python dictionaries with no dependencies on modern interpreters. Its default branch is a version branch, 1.3.x, its version lives only in pyproject.toml, and its testing instructions point at a file that is not in the tree.
- Who is it for?
- Cerberus is a reasonable choice for validating dictionaries against a schema when you want type checking without a dependency, and its extension model is built for adding your own rules. Before you rely on it, two things are worth settling.
- Can I use it commercially?
- Yes. ISC is a permissive licence: you can use, modify and sell software built on it, as long as you keep its copyright and licence notices.
- Is it still maintained?
- Yes. The repository last received commits 45 days ago.
- What is it written in?
- Mainly Python, according to GitHub's language statistics.
Answers come from the project's GitHub data, last synced on November 15, 2026, and from our analysis. They are not legal advice.
Star history
- pyeve/cerberus3,282 ★ · +2
| Repository | 2026-09-24 | 2026-11-15 |
|---|---|---|
| pyeve/cerberus | 3.3k | 3.3k |
2 Stars in the last 52 days, measured from the daily GitHub snapshot. Compare with other projects
Editorial analysis
The default branch is a version branch, and the links agree
The repository's default branch is not main or master. It is 1.3.x, and every internal link in the README is written against it: the contribution guidelines, the license file and the contribution guide all point at URLs containing the 1.3.x branch name. That choice has consequences for anyone who clones the repository. Development of the 1.3 line happens directly on the default branch, so a checkout is always the tip of that line rather than a stable point, and there is no separate development branch to track. The library itself follows semantic versioning from 1.2 onwards, where a major release sheds the old and defines a space for the new, minor releases add features and improvements, and micro releases polish. With no GitHub releases to read, that policy is the only statement about version stability the repository makes.
The test command names a file that is not here
The testing section gives two routes. The first is a single command, python setup.py test. The second is to install tox and run it, across all supported Python versions. Now look at what the repository contains: sixteen top level entries, among them pyproject.toml, tox.ini, MANIFEST.in, LICENSE, AUTHORS, CHANGES.rst, UPGRADING.rst, docs, artwork and the cerberus package itself. There is no setup.py. The tox route is the one backed by a file in the tree, tox.ini, while the setup.py route depends on a script that is not in the repository and on a test command that recent packaging tools no longer provide. The README still recommends running the contributed test suite on your target system if you are running an interpreter it no longer tests, so a contributor following that advice is pushed toward the branch of tooling that is missing.
The version number exists in exactly one place
The project has no GitHub releases, so there is no tag, no changelog entry generated from a tag and no release page to check. The current version, 1.3.8, is written in pyproject.toml, and the README's version badge is not a static label: it is an image pointing at the PyPI version endpoint for cerberus, meaning the number a reader sees is whatever the package index currently serves rather than what this branch contains. Two people are named in the manifest: Nicola Iarocci as author, with a contact address, and Frank Sachsenheim as maintainer, with a different one. The README credits the project to Nicola Iarocci as an open source project and points at the license file for terms.
One conditional dependency, and only below Python 3.8
The README's claim is that it has no dependencies, and on any interpreter a new project would use that is close to literal. The dependency list in the manifest holds a single entry, importlib-metadata, and it carries an environment marker restricting it to Python versions below 3.8. So the floor and the conditional dependency meet at the same boundary: on 3.8 and above nothing is installed alongside the library, while on the oldest supported interpreter exactly one backport is. That floor is declared as 3.7 or newer, and the classifiers run from 3.7 through 3.14, with both the CPython and PyPy implementations listed. The package is also marked as a production and stable release, and its keywords put normalization alongside validation, schema, dictionaries and documents, which is a fair description of what the library does to the data it checks.
The test suite turns warnings into errors
The pytest configuration in pyproject.toml sets filterwarnings to a single entry, error, which means any warning raised during the test run fails it. For a library whose whole job is inspecting other people's data, that is a strict setting: a deprecation warning from the standard library is a test failure rather than noise in the output. The rest of the tool chain follows the same pattern of small explicit choices. Black is configured with skip-string-normalization set to true, so quote style in the source is left as the author wrote it rather than being rewritten by the formatter. There is a pre-commit configuration, a separate linting configuration file, a tox configuration for the multi-interpreter run, and a Read the Docs configuration for the documentation build.
Interpreter support is promised six months past end of life
The README states its support policy as a testing commitment rather than a compatibility claim. The intent is to test Cerberus against all CPython interpreters at least until half a year after their end of life, and against the most recent PyPy interpreter, as a requirement for making a release. That leaves a defined window in which a new interpreter is covered and a longer tail in which it is not. The advice for anyone still running an interpreter with a security problem in their setup is specific: use the latest minor version branch from the time when that interpreter was still being tested, where subsequent minor versions have good chances of working too, and in any case run the contributed test suite on the target system rather than trusting the version list.
A licence badge is defined and never placed
Two small details in the packaging show how carefully this repository is put together. The README defines four image substitutions, for the latest version, the Python versions, the black code style and the software licence, and only three of them appear in the visible text. The licence badge is defined with a PyPI licence endpoint as its target and then never used, while the copyright section simply links to the license file. The second detail is packaging: include-package-data is set to false, so data files inside the package are not collected automatically, which is why MANIFEST.in sits at the root, and the packages find directive is narrowed to include only the cerberus package. The rest of the tree is documentation and history, with docs, a Read the Docs configuration, CHANGES.rst, UPGRADING.rst and an artwork directory.
Editorial conclusion
Cerberus is a reasonable choice for validating dictionaries against a schema when you want type checking without a dependency, and its extension model is built for adding your own rules. Before you rely on it, two things are worth settling. The version you get is whatever PyPI serves, since the repository publishes no releases of its own and the current number, 1.3.8, lives only in pyproject.toml on the 1.3.x branch. And the test instructions in the README do not match the repository, because the documented command runs setup.py and there is no setup.py here. For running the suite, tox.ini is the file that is actually present.
Frequently asked questions
how to install cerberus
With pip install cerberus. There is no required dependency on modern interpreters: the manifest's single dependency, importlib-metadata, is restricted to Python versions below 3.8, and the library declares support for 3.7 or newer.
Which Python versions does Cerberus support?
The manifest sets requires-python to 3.7 or newer and lists classifiers from 3.7 through 3.14 for both CPython and PyPy. The README states the policy as testing all CPython interpreters until half a year after their end of life, plus the most recent PyPy.
How do I run the Cerberus test suite?
The README offers python setup.py test or tox. There is no setup.py among the repository's top level entries, while tox.ini is present, so the tox route is the one a file in this repository supports.
Who maintains Cerberus?
The manifest names Nicola Iarocci as the author and Frank Sachsenheim as the maintainer, each with a contact address. The README credits the project as an open source project by Nicola Iarocci and points to the license file for terms.
What license is Cerberus released under?
ISC. The manifest points its license field at the LICENSE file rather than spelling out terms, and its classifiers include the ISC license classifier alongside a production and stable development status.
Official sources
Add this badge to your README
If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.
[](https://hysenlabs.com/projects/pyeve-cerberus)Star history chart for your README
[](https://hysenlabs.com/projects/pyeve-cerberus)Was this useful?