Environment:Protectai Modelscan Python Core Runtime

Knowledge Sources	modelscan pyproject.toml
Domains	Infrastructure, Security
Last Updated	2026-02-14 12:00 GMT

Overview

Python 3.10-3.12 environment with Click, NumPy, Rich, and tomlkit for running the ModelScan CLI security scanning tool.

Description

This environment defines the core runtime for ModelScan, a tool that detects unsafe operations in serialized ML model files. It requires Python 3.10 or higher (up to 3.12) and a set of mandatory dependencies: Click for CLI handling, NumPy for binary format parsing, Rich for console reporting, and tomlkit for settings file management. No GPU or special hardware is required — ModelScan performs static analysis of model file bytecode, not model inference.

Usage

Use this environment for all ModelScan operations: CLI scanning, programmatic scanning, and settings file generation. This is the base requirement for every workflow in the repository. Optional dependencies (TensorFlow, h5py) extend scanning capabilities to additional model formats but are not required for pickle, PyTorch, or NumPy scanning.

System Requirements

Category	Requirement	Notes
OS	Linux, macOS, Windows	Cross-platform Python package
Hardware	Standard CPU	No GPU required; static analysis only
Disk	Minimal	Only needs space for model files being scanned

Dependencies

System Packages

No system-level packages required beyond a Python installation

Python Packages

`python` >= 3.10, < 3.13
`click` >= 8.1.3
`numpy` >= 1.24.3
`rich` >= 13.4.2, < 15.0.0
`tomlkit` >= 0.12.3, < 0.14.0
`setuptools` == 80.9.0

Credentials

No credentials or environment variables are required for core operation.

The following optional variable is used in test environments only:

`TF_USE_LEGACY_KERAS`: Set to `"1"` in tests to enable legacy Keras API compatibility with newer TensorFlow versions.

Quick Install

# Install modelscan with core dependencies only
pip install modelscan

# Or install from source
pip install click>=8.1.3 numpy>=1.24.3 "rich>=13.4.2,<15.0.0" "tomlkit>=0.12.3,<0.14.0"

Code Evidence

Python version constraint from `pyproject.toml:15`:

python = ">=3.10,<3.13"

Core dependency declarations from `pyproject.toml:16-19`:

click = "^8.1.3"
numpy = ">=1.24.3"
rich = ">=13.4.2,<15.0.0"
tomlkit = ">=0.12.3,<0.14.0"

Settings file auto-discovery using `os.getcwd()` from `cli.py:108-110`:

settings_file_path = Path(
    settings_file if settings_file else f"{os.getcwd()}/modelscan-settings.toml"
)

Common Errors

Error Message	Cause	Solution
`ModuleNotFoundError: No module named 'click'`	Core dependency not installed	`pip install click>=8.1.3`
`ModuleNotFoundError: No module named 'rich'`	Core dependency not installed	`pip install "rich>=13.4.2"`
`No settings file detected at .../modelscan-settings.toml. Using defaults.`	No custom settings file in CWD	Normal behavior; use `modelscan create-settings-file` to create one

Compatibility Notes

Python 3.12: Supported, but TensorFlow extras may have limited compatibility (see TODO in `pyproject.toml:23`: "Add py3.12 once TF release supports").
Python < 3.10: Not supported. The `>=3.10` constraint is enforced by the package metadata.
Python >= 3.13: Not supported. The `<3.13` upper bound is explicitly set.

Related Pages

Page Connections

Double-click a node to navigate. Hold to expand connections.

Principle

Implementation

Heuristic

Environment