This package provides a small, reusable validity frame that stays outside of other projects. It is designed to be installed once and imported anywhere without rewriting the validation logic.

• a direct Python API: ValidityFrame
• a helper to load a frame automatically from entry points: create_validity_frame()
• a minimal default implementation that you can extend or replace

It means the smallest working file structure needed to start the package, install it, and extend it later. It includes only the essential files: packaging metadata, a Python module, tests, and a short README.

Local editable installation:

pip install -e validity_frame

Minimal usage:

from validity_frame import ValidityFrame

vf = ValidityFrame(tolerant=True)
vf.verify(state, steps)

The package does not depend on other frameworks and only expects compatible objects through duck-typing.

The package registers a default entry point named default in the validity_frame.frames group. You can load it with:

from validity_frame import create_validity_frame

vf = create_validity_frame(tolerant=True)
vf.verify(state, steps)

If you later publish another implementation, you can register it in the same entry-point group and load it by name.

[project.entry-points."validity_frame.frames"]
my_custom_frame = "other_package.module:MyValidityFrame"

Then load it with:

from validity_frame import create_validity_frame

vf = create_validity_frame("my_custom_frame", tolerant=False)

An executable sequence diagram showing how the auto-capture integration operates is included in the package docs:

• validity_frame/docs/validity_frame_runtime_sequence.puml — shows adapter/entry-point resolution, collector start/end, probe collection, and how the runner writes the capture artifacts.

Render the diagram with PlantUML or view the original in the main project docs at docs/source/architecture/process/diagrams/validity_frame_runtime_sequence.puml.