🧠 LLMFuzzer - Fuzzing Framework for Large Language Models 🧠 LLMFuzzer is the first open-source fuzzing framework specifically designed for Large Language Models (LLMs), especially for their integrations in applications via LLM APIs. 🚀💥
An intentionally vulnerable AI chatbot to learn and practice AI Security.
Securing LLM's Against Top 10 OWASP Large Language Model Vulnerabilities 2024
Working Memory Attack on LLMs
OllaDeck is a purple technology stack for Generative AI (text modality) cybersecurity. It provides a comprehensive set of tools for both blue team and red team operations in the context of text-based generative AI.
OWASP GenAI LLM Top 10 2026 — gayriresmî Türkçe çeviri (CC BY-SA 4.0)
AI Red Team & Blue Team Tips & Tricks!
A high-performance string formatter written in Rust. This project detects and blocks LLM prompt injection and jailbreak attacks. It also features a customizable rule-based system and defends against obfuscated prompt attacks.
[COLM 2025] VisualTrap: A Stealthy Backdoor Attack on GUI Agents via Visual Grounding Manipulation
Default-deny execution gate for AI agents, APIs, and distributed systems. Policies, drift detection, hard invariants, post-quantum signed permits.Written in Rust. Says no unless policy says yes.
The Citadel is not just a training platform; it is a battleground. As AI systems integrate deeper into our critical infrastructure, the attack surface expands exponentially. This application is a purpose-built LLM Pentesting Environment designed to simulate real-world threats against Large Language Models.
AiShields is an open-source Artificial Intelligence Data Input and Output Sanitizer
A powerful, community-curated toolkit to attack, evaluate, defend, and monitor Large Language Models (LLMs) — covering everything from prompt injection to jailbreak detection.
Protect your AI interactions
Defend against prompt injection, jailbreaks, data leakage, and PII exposure in production applications.
A CLI security auditor that finds vulnerabilities unique to LLM-generated code, what SAST tools miss.
基于开源题库（六大关卡 95 题）+ 纯规则判分引擎（不使用 LLM-as-Judge）+ GitHub Actions CI 复跑 + BYOK 自带密钥试测页，接入任意 OpenAI 兼容 API → 自动完成「安全常识 → 越狱拒答与钓鱼识别 → 恶意代码判断 → 代码漏洞识别 → 密码学与编码 → 日志研判与应急响应」六关测评，产出安全评分与公开排行榜。
RAG pipeline red-teamed against prompt injection: poisoned document corpus, 20 attack payloads across 6 categories, LangGraph guard (sanitizer + LLM classifier + output validator). Baseline 40% attack success → 0% defended, +0.43s latency cost, honest false-positive analysis.
A series of serverless functions/resources (and Terraform) for consuming language model inputs and outputs to S3, enriching the data via sentiment analysis and topic modelling, loading to DynamoDB and subsequently monitoring for configurable deviation within the latent vector space.
A Python SDK (backed by Rust) for establishing cryptographic security boundaries in LLM prompts.
To associate your repository with the llmsecurity topic, visit your repo's landing page and select "manage topics."