Multi-Agent LLM System for Automated Vulnerability Discovery and Reproduction

Source: Hacker News

Abstract

Software vulnerabilities pose critical security threats, with nearly 50,000 CVEs reported in 2025. While Large Language Models (LLMs) show promise for automated vulnerability detection, three key challenges remain.

1. False positives & reproducibility – LLM‑generated vulnerability reports suffer from high false‑positive rates and lack reproducible verification.
2. Granularity of localization – Existing LLM‑based approaches use suboptimal granularities: function‑level analysis overlooks bugs when context becomes extensive, while line‑level analysis lacks sufficient context.
3. Complex reasoning – Difficulty reasoning about vulnerabilities with complex cross‑function dependencies and triggering conditions.

We present FuzzingBrain V2, a multi‑agent system that addresses these gaps through four key contributions:

1. Fully automated vulnerability analysis built on Google’s OSS‑Fuzz, ensuring all reported vulnerabilities are fuzzer‑reproducible.
2. Suspicious Point, a novel control‑flow‑based abstraction for precise vulnerability localization at the optimal granularity.
3. Logic‑driven hierarchical function analysis with dual‑layer fuzzing, enhancing function coverage under resource constraints.
4. MCP‑based static and dynamic analysis tools with context engineering, improving reasoning about complex vulnerabilities.

On the AIxCC 2025 Final Competition C/C++ dataset, FuzzingBrain V2 achieved a 90 % detection rate (36 of 40 vulnerabilities). In real‑world deployment, it discovered 29 zero‑day vulnerabilities across 12 open‑source projects, all confirmed and fixed by maintainers, with 2 assigned CVE IDs.