Why repo-level verification: single-function benchmarks miss real software
dawnsongtweets · x · 2026-08-23
Motivation for Vero: agents are impressive at writing code, but can we trust it at real software scale? Existing benchmarks target single functions or proof generation on a fixed implementation, whereas real verified software (OS kernels, crypto protocols, distributed systems) lives in multi-module repositories where code, specs, and proofs are deeply interdependent—a single refactor can invalidate proofs across the codebase. No prior benchmark evaluates joint code-and-proof generation and formal verification at this scale.
More from Research
- False Collatz Proof via a Lean Bug: Talk Probes Proof Assistants as LLM-Found Bug Targets — marc_lelarge · 2026-10-16
- Goodfire's Nick Cammarata argues neural networks are 'already clean' in new interpretability essay — ctjlewis · 2026-10-16
- NEJM AI: prompt tuning cuts LLM clinical test-order cost ratio to 1.4x across 19 systems — dvir_a · 2026-10-16
- Prompting cuts LLM over-testing costs across 19 systems, but diagnosis agreement drops — dvir_a · 2026-10-16
- AI Recommends 3.2x Costlier Care; Models Add Tests in 2/3 of Visits Physicians Ordered None — dvir_a · 2026-10-16
- 24 AI Systems Recommend $314 Per Visit in Care vs Physicians' $97, a 3.2x Gap — dvir_a · 2026-10-16