#1LLM security, prompt injection, model extraction, adversarial AI, and AI red teaming tools.
Kitploit recommended

Research code for a gray-box trojan attack that flips a single KV-cache bit in fine-tuned LLM classifiers and measures per-class attack success rate.

Reverse engineered Linux kernel driver and userspace library for the Apple Neural Engine (ANE), enabling hardware access and analysis on Linux…

SHAREM is a shellcode analysis framework, capable of emulating more than 45,000 WinAPIs and virutally all Windows syscalls. It also contains its own…

AI-native code security auditor on AgentField that proves exploitability with verdicts, traces, and actionable evidence.

Your agent is a security risk, so treat it like one. yoloAI does AI agent sandboxing right.

CVE-Factory

An Open Harness and Benchmark for AI in Cybersecurity Operations.

Mobile camera-based application that attempts to alter photos to preserve their utility to humans while making them unusable for facial recognition…

The code for ACM MM2024 (Multimodal Unlearnable Examples: Protecting Data against Multimodal Contrastive Learning)

[CVPR 2025-ADVML] Official Repository for `Attacking Attention of Foundation Models Effectively Disrupts Downstream Tasks`

Static analysis CLI that scans codebases for LLM prompt-injection, data-exfiltration, jailbreak, and unsafe agent/tool vulnerabilities. Runs fully…

How AI turns password attacks from blind enumeration into adaptive campaigns - and the detection rules and countermeasures that stop them. Research…

Proof-of-concept exploit for CVE-2025-32434, a pickle deserialization vulnerability in PyTorch's weights_only mode, demonstrating malicious tar file…

A lightweight mcp to prevent poisoning CVE (CVE-2025-54136), researchers hijacking Claude Code/Copilot/Gemini via prompt injection, and hundreds of…

The Governed Agentic AI Operating System — Rust + Tauri 2.0 | 65 crates, 658 commands, 84 pages, 5,029 tests, 10/10 OWASP

Read-only scanner for what lets a repository run code in a coding agent (Claude Code, Codex, Cursor, Copilot): git settings, hooks, and committed MCP…

Proof-of-concept demonstrating remote code execution via request-side prompt injection in OpenClaw Agent Platform, exploiting lack of integrity…

Sub-millisecond VM sandboxes for AI agents via copy-on-write forking