jazzmine-security
Moderation, sanitization, and toxicity detection primitives for securing AI input and output.
Moderation, sanitization, and toxicity detection for AI input and output: input/output moderators, HTML/CSV/PDF sanitizers, and configurable threshold-based toxicity blocking for Python services.
Quickstart scope
Start with the minimal end-to-end usage flow, then add sanitizer variants and reference pages as needed for production hardening.
Install
python3 -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade pip
pip install jazzmine-securityOptional GPU-enabled PyTorch install:
pip install torch --index-url https://download.pytorch.org/whl/cu121Key features
- Input and output moderation with label and confidence scoring.
- HTML sanitization for model output before client delivery.
- Toxicity detector with configurable threshold-based blocking.
- CSV and PDF sanitizer variants for non-HTML content pipelines.
Reference
Use the sidebar for detailed security internals — moderators, sanitizers, and detector internals.