Sherlock has officially launched Sherlock Audit Engine, a security auditing platform it had been quietly testing with protocol teams. The system coordinates Frontier LLMs, purpose-built AI auditors and AI-enabled security researchers, all analyzing the same codebase. Findings are judged, validated, deduplicated and consolidated into a single final report. In June, the audit of Polygon PoS consensus client Heimdall V2 served as a proving ground for the model, demonstrating that no single approach captures the full security picture. Audit Engine is designed to incorporate new models and methodologies as they emerge, while providing protocols with a consistent environment to measure actual performance against their code.
Source: Read the original article
