# Content Guard

Content Guard is a CLI tool and Python library for detecting harmful or inappropriate text generated by AI models. It uses pre-trained NLP models to classify content into categories like hate speech, toxicity, or explicit material, helping developers filter problematic outputs effectively.

## Features
- Detects harmful or inappropriate text using pre-trained NLP models.
- Classifies content into categories such as hate speech, toxicity, or explicit material.
- Outputs flagged content as JSON.

## Installation

Install the required dependencies:

```bash
pip install transformers pytest
```

## Usage

### CLI

To use Content Guard as a CLI tool:

```bash
python content_guard.py --input <input_file_path> --output <output_file_path>
```

- `--input` or `-i`: Path to the input text file.
- `--output` or `-o`: Path to save the flagged content as JSON. If not provided, the flagged content will be printed to the console.

### Python Library

You can also use Content Guard as a Python library:

```python
from content_guard import classify_text
from transformers import pipeline

classifier = pipeline('text-classification', model='unitary/toxic-bert')
text = "This is a toxic comment."
flagged = classify_text(text, classifier)
print(flagged)
```

## Testing

To run the tests:

```bash
pytest test_content_guard.py
```

## License

MIT License