datasheet-ci Python Validator

Validate one local Datasheet, Model Card, or Data Card and receive structured missing-section and PII-pattern evidence.

This Python package is for authors who want the same required-heading lists used by AuraOne Datasheet CI before opening a pull request.

Inspectable Output

The CLI writes JSON with ok, missing, and warning-only piiWarnings for the supplied file.

Runtime Boundary

The CLI reads one local Markdown file and performs local regular-expression checks. It makes no network requests, uploads no document content, and does not determine whether a matched string is actually personal data.

Install and Quickstart

From the datasheet-ci repository root:

python -m pip install ./python_validator
datasheet-ci examples/valid_datasheet.md
datasheet-ci examples/invalid_datasheet.md
datasheet-ci path/to/model-card.md --kind model_card

Release Status

Verified July 13, 2026: datasheet-ci==0.2.1 is published on PyPI and the source is included in the dedicated v0.2.1 repository release.

Limits

This validator checks completeness signals, not factual accuracy, privacy compliance, or documentation quality.

Next Action

Install the validator from python_validator/, run it on the document you plan to submit, add every missing heading, and review each PII-like match manually before opening the pull request.