datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
shell-attack-evolution-dataset
Shell Honeypot Attack Request–Response Dataset
A standardized, MITRE ATT&CK–annotated dataset of post-login shell
attacks captured by Cowrie SSH/Telnet
honeypots across two collection periods — 2021–2022 and 2024. It pairs
attacker shell commands with real captured system responses, enabling both
longitudinal threat analysis and the training/evaluation of AI-driven honeypots.
This is the open-source release accompanying the paper “Unveiling Evolving
Threats: A Data Analysis… See the full description on the dataset page: https://huggingface.co/datasets/zyw-286/shell-attack-evolution-dataset.task325_jigsaw_classification_identity_attack
Dataset Card for Natural Instructions (https://github.com/allenai/natural-instructions) Task: task325_jigsaw_classification_identity_attack
Additional Information
Citation Information
The following paper introduces the corpus in detail. If you use the corpus in published work, please cite it:
@misc{wang2022supernaturalinstructionsgeneralizationdeclarativeinstructions,
title={Super-NaturalInstructions: Generalization via Declarative Instructions on 1600+… See the full description on the dataset page: https://huggingface.co/datasets/Lots-of-LoRAs/task325_jigsaw_classification_identity_attack.attackqa
AttackQA: Development and Adoption of a Dataset for Assisting Cybersecurity Operations using Fine-tuned and Open-Source LLMs
license: apache-2.0
This dataset is derived from the MITRE ATT&CK® knowledge base that bears the following license:
© 2025 The MITRE Corporation. This work is reproduced and distributed with the permission of The MITRE Corporation.
In using the dataset, please consider citing the following paper:
misc{c:attackqa,
title={AttackQA: Development… See the full description on the dataset page: https://huggingface.co/datasets/sambanovasystems/attackqa.CoER-Attacker-SFT
CoER Attacker SFT
Project page · Paper · Code
Stage 1 supervision for initializing the adaptive attacker. Successful conversations retain all attacker turns, including earlier attempts that provide context for later adaptation.
Contents
Split
File
Size
train
train.jsonl
3,995 conversations
The corpus contains 11,655 assistant/attacker turns. Preserve all assistant-turn supervision; do not reduce a conversation to its final payload.
Load… See the full description on the dataset page: https://huggingface.co/datasets/Z-Edgar/CoER-Attacker-SFT.elite-attack
Elite Attack Dataset
A collection of 100 prompt injection test cases for evaluating LLM security defenses.
Dataset Description
This dataset contains prompt injection attacks designed to test the security of Large Language Models. Each attack scenario includes system messages with embedded secrets that the attacks attempt to extract.
Dataset Structure
100 test cases across 10 attack families
Verified effectiveness against Llama-3.2-3B-Instruct
Diverse attack… See the full description on the dataset page: https://huggingface.co/datasets/zyushg/elite-attack.geometry-of-harmfulness-in-multi-turn-attacks
Geometry of Harmfulness — Multi-Turn Attack Conversations
Raw multi-turn attack conversations accompanying the paper
The Geometry of Harmfulness in Multi-Turn Attacks. These are the conversations
from which the paper's hidden-state representations are extracted; the analysis
code lives in the companion repository.
Conversations were generated by running three multi-turn attack frameworks —
Crescendo, ActorAttack, and X-Teaming (attacker & judge: GPT-4o) —
against three… See the full description on the dataset page: https://huggingface.co/datasets/yelyzavetahusieva/geometry-of-harmfulness-in-multi-turn-attacks.attack-mitigation-corpus-v1
Attack-Mitigation-Corpus v1
One JSONL file of 123 security records. Each record is either a described attack
technique or a mitigation, and each attack record is paired with a mitigation record
covering a related defensive concern.
Measured composition
metric
value
command
rows
123
wc -l < attack_mitigation_corpus.jsonl
kind == "attack"
106
python3 -c "import json,collections;print(collections.Counter(json.loads(l)['kind'] for l in… See the full description on the dataset page: https://huggingface.co/datasets/CatQualia/attack-mitigation-corpus-v1.attackAdema_ATTACK_DS
Adema ChatML Dataset
This dataset is a reformatted version of the original Adema research dataset, structured specifically in ChatML format for seamless integration with SFTTrainer and ChatML-based instruct models.
Dataset has three splits:
Train: 80%
Validation: 10%
Test: 10%
Dataset Details
Original Paper: Automated discovery and mapping ATT&CK tactics and techniques for unstructured cyber threat intelligence
Format: ChatML (messages column containing role and… See the full description on the dataset page: https://huggingface.co/datasets/madox81/Adema_ATTACK_DS.llm-attacked-prompts-clm
LLM-Paraphrased Adversarial Prompts
LLM-paraphrased adversarial prompts for three code-generation benchmarks
(MBPP+, HumanEval+, CanItEdit), used by
RobustEval-CLM's
LLMParaphraseAttack.
Each row corresponds to one task in the source benchmark and carries the
original prompt alongside an adversarial rewrite produced by an LLM under a
BERTScore faithfulness constraint.
Configs
config
source benchmark
rewrite surface
mbpp
MBPP+
line 1 of the 4-line prompt… See the full description on the dataset page: https://huggingface.co/datasets/TheFatBlue/llm-attacked-prompts-clm.attacker-zero-windows-v1
Attacker Zero Windows v1
This dataset is a prescored local-window derivative of
OpAI-Bench1/OpAI-Bench for the
attacker-zero Verifiers environment.
Each row contains one human / AI-aided / human sentence window from OpAI-Bench:
[Previous]: human sentence
[TARGET]: AI-aided sentence
[Next]: human sentence
The dataset intentionally stores raw window fields and deterministic detector
scores, not prompts. The environment owns prompt rendering, action formatting,
turn logic, and… See the full description on the dataset page: https://huggingface.co/datasets/oliveirabruno01/attacker-zero-windows-v1.CICIoMT2024_Attacks_Orion_v0.1.0_0x0rr-circuit-breakers-attack-completions
RR (Circuit Breakers) attack completions with three-judge scores
This dataset bundles attack completions generated against
GraySwanAI/Llama-3-8B-Instruct-RR
(the "circuit breakers" defense), each scored by three independent judges:
local:strongreject (Lin et al., StrongREJECT classifier — most permissive)
local:harmbench (HarmBench classifier — middle)
local:gpt_oss (gpt-oss-safeguard-20b — strictest)
Headline finding: judges DISAGREE dramatically on… See the full description on the dataset page: https://huggingface.co/datasets/samuelsimko/rr-circuit-breakers-attack-completions.m2s-safemt-attack
M2S → SafeMT Attack_600 (single-turn)
This dataset contains single-turn prompts produced by applying the M2S conversion (“Hyphenize / Numberize / Pythonize”) to multi-turn jailbreak dialogues referenced in the M2S paper. We retain only minimal fields (IDs + prompt text); no model outputs or labels are included.
Intended use
Research on jailbreak robustness and defenses; please handle responsibly (adversarial/harmful prompts). We recommend enabling gated access.… See the full description on the dataset page: https://huggingface.co/datasets/lvogel123/m2s-safemt-attack.