#1Security papers, research projects, technical references, and reproducible material for defensive or authorized analysis.
Kitploit recommended

my personal exploit of CVE-2022-0847(dirty pipe)
CVE-2026-25262 (Qualcomm BootROM) deep-dive on the Black Hat Asia 2026 'A journey through Sahara' talk + read-only Sahara/EDL recon harness. GPLv3.

A serverless networking protocol designed for resilient state synchronization between autonomous agents in fragmented, low-bandwidth networks

Robust audio watermarking framework embedding binary messages into magnitude spectrograms, with differentiable attack simulation, Q-Former pooling,…

Defensive framework that maintains a safety-focused shadow memory to detect and block prompt-injection and long-horizon threats against LLM agents…

Proof-of-concept exploit code for Linux kernel use-after-free vulnerabilities (CVE-2022-2585, CVE-2022-2586, CVE-2022-2588) enabling local privilege…

Temporary Root Research on Poco M7 Plus (SM6375) via Qualcomm GBL Exploit (CVE-2026-24088) + GhostLock Kernel Analysis (CVE-2026-43499)

Proof of concept and technical write-up for CVE-2026-56096, a blind Solr query injection in TYPO3 EXT:solr enabling unauthenticated field enumeration…

Google patched CVE-2025-10585, a Chrome V8 zero-day under active exploitation — here’s what it is, why it matters, and how to stay safe.

Informe técnico de una vulnerabilidad ya corregida en Instagram Notes que exponía el audio original de vídeos mediante URLs directas.

Tracker of publicly reported prompt-injection techniques, broken down by delivery method, encoding, and propagation behavior, with confirmed models,…

罗技云掌机 · GhostLock CVE-2026-43499 root 尝试

Research demonstration of indirect prompt injection attacks to control autonomous LLM-based web agents, with tools for trigger optimization and…

Reverse engineering write-up of Python shellcode that APC-injects into AnyDesk, exfiltrates to a C2 over HTTPS with AES/RSA, and persists via…

Exploitation framework for CVE-2026-86950, a CoreGraphics integer overflow: crafts malicious TrueType fonts and PDFs, extracts ASLR leaks, and builds…

Defender framework for LLM agent security that compiles task contracts, validates capability manifests, and checks effects via PLANT/WRAP proof…

Runtime guard skill that detects and blocks malicious skill-induced attacks in skill-augmented agents, with an MCTS loop that evolves the guard…

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