Tools

phil-fan/fs-attack

Implementation of the paper "Rethinking Latency Denial-of-Service: Attack the LLM Serving Framework, Not the Model"

stardust457/sldr

Defense framework that mitigates malicious fine-tuning of LLMs using selective layer recovery and dynamic routing, with training, harmfulness…

uiuc-kang-lab/persistbd

Research code for training adapters that make LLM agent backdoors survive benign fine-tuning, with trigger-rate scoring, SWE-bench evaluation, and…

deep-jlu/skillpoison

Research pipeline that constructs verified successful experiences and organizes them into formation records to progressively poison model skills via…

zju-llm-safety/ascent

Fine-tuning framework that applies first-order optimal safety calibration and periodic recalibration to LLMs, preserving safety-compatible updates…

abhinavdubey30/perturbed-embedding-vectors

Code, response logs and labels for the paper Jailbreaking Open-Weight LLMs via Random Embedding Perturbations

jisung-pacific/hdi-graphrag-attack

Research implementation of Hop-Decayed Influence (HDI) and the 3S attack framework, exposing structural auxiliary indexing vulnerabilities in…

greshake/llm-security

Proof-of-concept demos and research on indirect prompt injection attacks against application-integrated LLMs, covering data exfiltration, remote…

euph00/zoneclaw-code

Benchmark and defense code for persistent memory attacks on OpenClaw-style computer-use agents, with memory-zoning mitigation, attack scenarios, and…

sliu11-byte/mea-bench

A lifecycle benchmark for black-box LLM extraction attacks, defenses, and adaptive attacks.

js-lee-ai/refusal-relocates

Refusal localizes, the damage relocates, safety layers under few-sample fine-tuning

kevinchunye/prompt_injection

Benchmark harness measuring where prompt injection defenses fire in tool-using LLM agent pipelines, tracking canary tokens across exposed, persisted,…

facebookresearch/meta_secalign

Open-weight LLMs and training/evaluation code for defending against prompt injection attacks, with benchmarks for agentic tool-calling and…

kaill-jlq/mmskillrisk

Benchmark and evaluation harness testing whether LLM agents resist malicious instructions hidden in multimodal skill images, with 108 cases across…

zju-llm-safety/trace

Agentic jailbreak framework for LLM-based agents using scheme-based task decomposition, multi-turn disguising strategies, and adaptive self-evolution…

zju-llm-safety/darwin

Evolutionary LLM jailbreak and guardrail framework that grows a reusable strategy pool via genetic mutation, Markov selection, and online adversarial…

albert-y1n/piforge

Open framework for RL-based prompt injection red teaming, with a shared trainer, curriculum learning, and benchmarks like AgentDojo, InjecAgent, and…