Jailbreak Evaluation Framework (JEF)
JEF is a CVSS-like scoring system for quantifying LLM jailbreak tactics. It provides a structured, repeatable framework for stack-ranking jailbreak methods based on their real-world efficacy and risk, scoring across three dimensions:
- Model Impact Radius — How many models and vendors are affected
- Retargetability — How flexibly the tactic applies across content categories
- Output Fidelity — How accurate and detailed the resulting outputs are
Installation
pip install 0din-jef
Quick Start
from jef import tiananmen, meth, nerve_agent, harry_potter
# Score against standardized benchmarks (0-100)
result = tiananmen.score("LLM response text")
result = meth.score("LLM response text")
result = nerve_agent.score("LLM response text")
result = harry_potter.score("LLM output", "reference text")
# Compute composite JEF score (0-10)
from jef import calculator
jef_score = calculator(num_vendors=3, num_models=7, num_subjects=2, scores=[80, 75])
Documentation
For the full framework methodology, scoring algorithm, complete usage guide, and API reference, visit the JEF Documentation.
Resources
- Blog: Quantifying the Unruly — A Scoring System for Jailbreak Tactics
- Overview: Jailbreak Evaluation Framework
- JEF Calculator
- Standardized Testing (0DIN Researcher Authentication Required)
Releases
Releases are managed through GitHub Releases and automatically published to PyPI.
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Filter files by name, interpreter, ABI, and platform.
If you're not sure about the file name format, learn more about wheel file names.
Copy a direct link to the current filters
File details
Details for the file 0din_jef-0.8.1.tar.gz.
File metadata
- Download URL: 0din_jef-0.8.1.tar.gz
- Upload date:
- Size: 108.6 kB
- Tags: Source
- Uploaded using Trusted Publishing? No
- Uploaded via:
twine/6.1.0 CPython/3.13.13
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
0e69fde8ac646d9392cce47ead1775893f3dfa4dcd93cd34089cf163b44e76f5
|
|
| MD5 |
9328d61a2c328fb9ad84d1f49ba7b459
|
|
| BLAKE2b-256 |
b26465624739173d0b98710538273ea7ed5efff0cf6a78ebacbc81f3cad6f6a6
|
File details
Details for the file 0din_jef-0.8.1-py3-none-any.whl.
File metadata
- Download URL: 0din_jef-0.8.1-py3-none-any.whl
- Upload date:
- Size: 87.7 kB
- Tags: Python 3
- Uploaded using Trusted Publishing? No
- Uploaded via:
twine/6.1.0 CPython/3.13.13
File hashes
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 |
0d1ea2441e2eea7c5c5575003a79ad6b4fd6a669bcb1a43484048735b6b4cf8c
|
|
| MD5 |
d1e3b0f1838aa41e8ce37bc4ab15b2ab
|
|
| BLAKE2b-256 |
f8b72da77e56faffb36dec39dbdee5c048cef0ef6d0d417c542bc626c28bf4d3
|
