QuickProof: formal verification for C with specs written as plain C tests
thomasahle · x · 2026-11-04
Formal verification has been blocked by two barriers: specs had to be written in proof-assistant languages like Lean/Rocq, and proofs were often far longer than the code. GenProof's new QuickProof attacks the first barrier for C: you write the specification as a small C program that calls your library, much like a property-based test.
- AI agents now write the formal proofs; a proof checker (Rocq) verifies every step, so wrong proofs can't slip through.
- The subtle risk remains that an agent weakens the stated theorem or adds assumptions — but with specs in C itself, you can always read what you're proving.
- The post walks through proving properties of production C code (e.g. a compress/decompress round-trip property using nondet inputs and qpassert) and claims the same process finds bugs even in already-verified libraries.
More from coding & agent
- Concurrency limits are the hidden tax in voice AI — how Phonic solved it — graceisford · 2026-11-04
- Why AI agents optimize Go code better than Rust: domain expertise still steers the loop — unixterminal · 2026-11-04
- Alibaba's SPAR sales agent lifts conversion 15.8% in A/B test across 200k+ merchants — ycombinator · 2026-11-04
- 2637 prompts, 12.5B tokens, 383 PRs: designer opens up his entire AI-built portfolio process — tomjohndesign · 2026-11-04
- DeepSeek article outlines super-agent architecture: continuous, productive, proactive, persistent — LangChain · 2026-11-04
- Salesforce's 700-person AI org switched from Cursor to Claude Code, no mandate needed — aakashgupta · 2026-11-04