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

Educational Python target range simulating CVE-2026-22807, an AI supply chain RCE via TOCTOU in model loading. Includes vulnerable library, PoC…
TP Seguridad Informática UTN FRBA 2021

Scans Shodan for publicly exposed Ollama and llama.cpp servers, enumerating hosts, locations, and available models for security research.

AI-based malicious code detection and obfuscation system with microservices for detection, packing, and code confusion, plus user and admin services.

Exploits for CVE-2024-14027 creating using AI/Claude as a test.

OpenMed < 1.5.2 unauthenticated RCE via PII privacy-filter model loading and trust_remote_code=True

Public advisory landing page for CVE-2026-54519: missing ownership checks in ai-agent-automation memory APIs enabling cross-user memory read and…

CVE-2026-45033 PoC for Claude Code, not Github Copilot. Worked for Haiku 4.5.

An audit-oriented risk model for how agent-generated inferences can change epistemic status, gain operational authority, and influence later…

OWASP guide defining a standardized penetration testing methodology for humanoid, industrial, and autonomous robots, covering ROS/ROS2, DDS, CAN bus,…

Proof-of-concept demonstrating prompt injection in Langchain's GraphCypherQAChain leading to SQL injection in Neo4j databases. Includes Docker-based…

OWASP reference project for agentic AI security evidence: correlates agent execution, identity, tool calls, and authorization decisions into…

Claude Code and Codex skill for finding potentially vulnerable Brotli copies and guiding updates for CVE-2026-111497.

Proof-of-concept exploit for CVE-2026-33017, an unauthenticated RCE in Langflow's build_public_tmp endpoint, injecting a malicious custom component…

YARA-inspired prompt pattern matching engine that combines keyword, regex, semantic similarity, and LLM evaluation rules to detect jailbreaks and…

Agentic harness that automates vulnerability discovery and validation, orchestrating AI-driven analysis workflows to confirm exploitable findings.

Sentinel — DeanWay Labs' inbound governor. Guards what an AI agent hears: world content is data, never orders. The mirror of QuietOS.

Security research on context compaction in coding agents: shows how forged role labels in untrusted content survive summarization and get executed,…