Implementation:Protectai Modelscan ConsoleReport Generate

Knowledge Sources	ModelScan
Domains	ML_Security, Reporting
Last Updated	2026-02-14 12:00 GMT

Overview

Concrete tool for generating formatted scan reports (console and JSON) provided by the modelscan reports module.

Description

The ConsoleReport and JSONReport classes implement the Report abstract base class to produce scan output in two built-in formats. ConsoleReport uses the rich library to print color-coded, severity-grouped issue lists to the terminal. JSONReport serializes the complete results dictionary to JSON, optionally writing to a file. Both are loaded dynamically by ModelScan.generate_report() via the reporting.module setting.

Usage

Use these report classes when you need to:

• Display interactive scan results in a terminal (ConsoleReport)
• Export scan results as structured JSON for pipeline processing (JSONReport)
• Write scan results to a file for audit purposes (JSONReport with output_file)
• Build a custom report class following the same interface

Code Reference

Source Location

• Repository: modelscan
• File: modelscan/reports.py
• Lines: L14-98

Signature

class Report(metaclass=abc.ABCMeta):
    @staticmethod
    def generate(
        scan: ModelScan,
        settings: Dict[str, Any] = {},
    ) -> Optional[str]:
        """Abstract method. Generate report for completed scan."""

class ConsoleReport(Report):
    @staticmethod
    def generate(
        scan: ModelScan,
        settings: Dict[str, Any] = {},
    ) -> None:
        """
        Print rich-formatted console report.

        Args:
            scan: Completed ModelScan instance with issues, errors, skipped.
            settings: Report settings dict. Keys:
                - show_skipped (bool): Include skipped files in output.
        """

class JSONReport(Report):
    @staticmethod
    def generate(
        scan: ModelScan,
        settings: Dict[str, Any] = {},
    ) -> None:
        """
        Print JSON report to stdout and optionally to file.

        Args:
            scan: Completed ModelScan instance.
            settings: Report settings dict. Keys:
                - show_skipped (bool): Include skipped files in output.
                - output_file (str): Optional file path to write JSON report.
        """

Import

from modelscan.reports import ConsoleReport, JSONReport

I/O Contract

Inputs

Name	Type	Required	Description
scan	ModelScan	Yes	Completed ModelScan instance with populated issues, errors, skipped, and scanned attributes
settings	Dict[str, Any]	No	Report-specific settings: show_skipped (bool), output_file (str, JSONReport only)

Outputs

Name	Type	Description
ConsoleReport output	stdout	Rich-formatted text printed to terminal: summary section with issue counts by severity, issues grouped by severity with details, errors section, and optionally skipped files
JSONReport output	stdout + optional file	JSON-serialized results dict printed to stdout. If output_file is set, also written to the specified file path

Usage Examples

Console Report via ModelScan

from modelscan.modelscan import ModelScan

scanner = ModelScan()
scanner.scan("/path/to/model.pkl")

# Uses ConsoleReport by default
scanner.generate_report()

JSON Report with File Output

from modelscan.modelscan import ModelScan

scanner = ModelScan()
scanner.scan("/path/to/model.pkl")

# Switch to JSON reporting with file output
scanner._settings["reporting"]["module"] = "modelscan.reports.JSONReport"
scanner._settings["reporting"]["settings"] = {
    "show_skipped": True,
    "output_file": "/tmp/scan_report.json",
}

scanner.generate_report()

Direct ConsoleReport Usage

from modelscan.modelscan import ModelScan
from modelscan.reports import ConsoleReport

scanner = ModelScan()
scanner.scan("/path/to/model.pkl")

# Call ConsoleReport directly
ConsoleReport.generate(scan=scanner, settings={"show_skipped": False})

Related Pages

Implements Principle

• Principle:Protectai_Modelscan_Scan_Report_Generation

Requires Environment

• Environment:Protectai_Modelscan_Python_Core_Runtime