Metadata-Version: 2.4
Name: roguard-child-safety
Version: 5.0.0
Summary: Romanian and Ukrainian AI interaction testing, permission controls and guardrails
Author: RoGuard contributors
License-Expression: MIT
Project-URL: Homepage, https://github.com/xyz-rafael-xyz/roguard-child-safety
Project-URL: Repository, https://github.com/xyz-rafael-xyz/roguard-child-safety
Project-URL: Documentation, https://github.com/xyz-rafael-xyz/roguard-child-safety/blob/main/README.md
Project-URL: Issues, https://github.com/xyz-rafael-xyz/roguard-child-safety/issues
Keywords: child-safety,Romanian,Ukrainian,evaluation,guardrails
Classifier: Development Status :: 4 - Beta
Classifier: Intended Audience :: Developers
Classifier: Programming Language :: Python :: 3
Classifier: Programming Language :: Python :: 3.9
Classifier: Programming Language :: Python :: 3.10
Classifier: Programming Language :: Python :: 3.11
Classifier: Programming Language :: Python :: 3.12
Classifier: Programming Language :: Python :: 3.13
Classifier: Programming Language :: Python :: 3.14
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
Requires-Python: >=3.9
Description-Content-Type: text/markdown
License-File: LICENSE
Provides-Extra: semantic
Requires-Dist: torch<3,>=2.6; extra == "semantic"
Requires-Dist: transformers<5,>=4.57; extra == "semantic"
Requires-Dist: numpy<3,>=1.26; extra == "semantic"
Provides-Extra: train
Requires-Dist: numpy>=1.26; extra == "train"
Requires-Dist: pyyaml>=6; extra == "train"
Requires-Dist: torch>=2.3; extra == "train"
Requires-Dist: transformers<5,>=4.46; extra == "train"
Requires-Dist: peft<1,>=0.13; extra == "train"
Requires-Dist: accelerate<2,>=1; extra == "train"
Provides-Extra: mlx
Requires-Dist: mlx-lm<0.32,>=0.31.3; extra == "mlx"
Provides-Extra: test
Requires-Dist: pytest>=8; extra == "test"
Requires-Dist: jsonschema<5,>=4.23; extra == "test"
Provides-Extra: message-train
Requires-Dist: numpy<3,>=1.26; extra == "message-train"
Requires-Dist: scikit-learn<2,>=1.5; extra == "message-train"
Dynamic: license-file

# RoGuard — Romanian and Ukrainian interaction testing and guardrails

RoGuard 5 provides a local bilingual laboratory for fictional AI interactions, source-linked permissions, minimum necessary disclosure, decision consistency and useful-task preservation. It includes an optional semantic text classifier and the existing declared-contract checks.

## Install and open

```sh
pip install --upgrade roguard-child-safety
roguard-studio --lab --open
```

The core package has no runtime dependencies and supports Python 3.9–3.14. The browser and CLI simulate actions; they do not contact guardians or execute external tools. A separately configured Python gateway can call trusted host callbacks after live policy and approval checks.

## What you can do

- Replay branching fictional scenarios against programmed references, recorded responses or your own explicit Python adapter.
- Identify the first failing step and reduce a scenario with a bounded replay budget.
- Retain source restrictions through summaries, proposed memory and tool results.
- Compute the exact minimum declared disclosure cost while covering required task fields.
- Test decision invariance and required policy transitions, retaining uncertainty.
- Compare permitted-task completion, unnecessary refusal and unauthorized proposals/deliveries.

```sh
roguard-lab demo --language uk --family revocation --target stale
roguard-lab compare --language ro
roguard-lab suite --fail-on-failure
```

For the stronger optional text classifier, use Python 3.10+ (tested on 3.12):

```sh
pip install --upgrade 'roguard-child-safety[semantic]'
roguard-model fetch
roguard-studio --conversation --open --backend semantic
```

The pinned model download is approximately 1.2 GB; subsequent inference is local. V5 does not change the v4 classifiers. Their synthetic results do not establish real-world child-language accuracy. Laboratory reference scores demonstrate controlled software behavior, not AI reasoning or human benefit. Provenance, coverage, costs and authority are host declarations; callback code is trusted, not sandboxed. Reduced scenario exports contain their input text.

[GitHub and full README](https://github.com/xyz-rafael-xyz/roguard-child-safety) · [Interaction guide](https://github.com/xyz-rafael-xyz/roguard-child-safety/blob/main/docs/INTERACTION_LAB.md) · [v5 evidence](https://github.com/xyz-rafael-xyz/roguard-child-safety/blob/main/docs/V5_RELEASE.md) · [Classifier evidence](https://github.com/xyz-rafael-xyz/roguard-child-safety/blob/main/docs/V4_RELEASE.md)
