


The fastest LoongArch sandbox


Research code for red-teaming AI auto-mode monitors, including simulation evals, fuzzing, and monitor implementations for Claude Code and Codex…

Original CVEs, exploit PoCs, and security advisories with detailed vulnerability chains, privilege escalation, and container escape techniques for…

An open testing platform that probes HTTP/1.1 servers against RFC 9110/9112 requirements, smuggling vectors, and malformed input handling. Add your…

A secure* runtime for autonomous AI agents. Policy from plain-English constitutions. (*https://ironcurtain.dev)

LAVA: Large-scale Automated Vulnerability Addition

A GPU-oriented coverage-guided fuzzer for userland CUDA applications

Docker lab reproducing CVE-2026-42533, a pre-auth nginx heap overflow and info leak via two-pass capture clobbering, with PoC scripts and patched…

Some setup scripts for security research tools.

Fully dockerized Linux kernel debugging environment

Automated reconnaissance framework with 17+ modules for subdomain enumeration, directory brute-forcing, JS/link mining, WAF fingerprinting, and…


ExploitGym is a large-scale, realistic benchmark built from real-world vulnerabilities designed to evaluate AI agents' ability to develop exploits.

Build and query a graph database representation of source code