For anyone relying on a smart-contract audit before holding assets, the useful question is what a tool might miss—not just what it finds. In an article published on June 5, 2026, the authors of solidity-auditor v3 reported 82.4% vulnerability recall in a DoDo competition case. The article says v3 led the open-source tools it compared on recall, with a run time of about 22 minutes and token use of roughly 3–5 million.
The article describes two changes behind v3: a shared reasoning discipline drawing on Feynman, Socratic and inversion methods, and three “gap hunter” agents focused on flow, numerical and trust gaps. It also provides installation and usage instructions.
These are results reported by the article, not a guarantee that the tool will catch every vulnerability. The authors explicitly caution that AI analysis cannot ensure complete security. For readers assessing a contract they may depend on, the reported recall is a reason to examine the tool’s findings and omissions—not to treat an automated audit as proof of safety.

