TL;DR: A new MIT-licensed GitHub repo indexes recent LLM vulnerability detection papers, projects, and agent workflows.
Summary: The 'Awesome Large Language Models for Vulnerability Detection' repo curates a table of 2025+ papers with links to GitHub implementations, plus an archive for older work. It also lists practical LLM-driven scanning, pentesting, and discovery tools, and references agent security workflows for Codex, Claude Code, and Cloudflare's parallel audit agents. The project is open-source under MIT.
Why it matters: It gives AI builders a single starting point for turning scattered LLM security research into working implementations. Watch the repo for emerging agent-based vulnerability detection patterns, and use its research-to-code links before building your own tooling.
Source: x_com