Heads up This site is currently under heavy development.
← all datasets
◆ Evals & benchmarks · Prompt injection & jailbreaks

HarmBench

MIT

A standardised red-teaming evaluation of harmful-behaviour refusals across attack methods and models.

Summary

Automated red teaming holds substantial promise for uncovering and mitigating the risks associated with the malicious use of large language models (LLMs), yet the field lacks a standardized evaluation framework to rigorously assess these methods. To address this issue, we introduce HarmBench, a standardized evaluation framework for automated red teaming. We identify key considerations previously unaccounted for in red teaming evaluations and systematically design HarmBench to meet these criteria.
from the project README
Formats
csv
License
MIT
Added
2026-09-10

Get it

From the project’s own instructions where it documents any; otherwise a plain clone.

  1. shell
    $ git clone https://github.com/centerforaisafety/HarmBench.git
  2. shell
    $ cd HarmBench && pip install -r requirements.txt && python -m spacy download en_core_web_sm

Contents

429 files 103.6 MB repository

  • .py 123
  • .png 113
  • .pkl 100
  • .yaml 34
  • .json 10
  • .csv 9
  • .ipynb 9
  • .md 9

Use it with

Commands are curated, not yet run by us.

Promptfoo — Run your promptfoo config against the dataset's rows as test cases.

sample.csv
$ promptfoo eval -c promptfooconfig.yaml -t sample.csv

DuckDB — Query the file in place with SQL — DuckDB detects CSV, JSON and JSONL by extension.

sample.csv
$ duckdb -c 'SELECT * FROM '\''sample.csv'\'' LIMIT 10'
my-toolchain — 0 tools
paste an install list to detect your tools

A brew list, a Brewfile, requirements.txt, a Dockerfile — or just the product names, free-form. Nothing leaves your browser.

    browse all tools →

    Send us feedback