JailBreakDefense / __init__.py
kriti0608's picture
Upload __init__.py
654be98 verified
raw
history blame contribute delete
383 Bytes
# src/__init__.py
"""
SafetyRepairKit-style minimal jailbreak detection and repair utilities.
Modules:
- rules.py β†’ regex-based jailbreak / policy-violation patterns
- detector.py β†’ aggregates rule hits into a risk score 0–1
- repair.py β†’ simple repair strategy (safe fallback on high risk)
- pipeline.py β†’ wraps a generation model with safety+repair
"""