AproxResearch: AI-Powered Web3 Exploit Engine

AproxResearch is a full-stack AI security research platform built to accelerate Web3 vulnerability discovery and validation. The backend powers an exploit-intelligence pipeline that ingests smart contracts, analyzes control flow and state transitions, and maps findings against known exploit classes such as access-control bypass, token/accounting mismatch, and economic attack surfaces. The frontend at aproxresearch.org provides an interactive workspace where researchers can review AI-generated findings, track severity, and organize manual validation notes. What makes the project practical is the focus on analyst workflow: the system does not stop at generic “possible issue” outputs. It structures findings with reproducible context, highlights likely impact paths, and supports evidence-driven triage so security reviewers can move faster from detection to validation. The platform is designed for both individual researchers and teams who need consistent, explainable, and auditable vulnerability research. Our goal is to reduce time-to-insight in smart contract audits while improving quality and repeatability, combining AI-assisted detection with human expert verification for real-world exploit research.