Reproduce “Modern Software Engineering with Verification Agents: A Spec-First Architecture for Autonomous Synthesis, Verification IRs, and Organizational Scale” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Jason Hickey

The authors propose a spec-first architecture where engineers review formal contracts and agents generate implementations and proofs. Its Verification IR uses higher-order abstract syntax and linear resource typing. Prior compiler work used 484 lines of formal mathematics versus 12,000 and took 3× less time. A router theorem checks one repaired listing in x86lean, not hardware execution or a complete workflow.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.