Anthropic formalizes Fermat's Last Theorem in Lean: 12M+ lines, 29k theorems
doodlestein · x · 2026-09-06
Overshadowed by OpenAI's Astra launch the same day: Anthropic announced it had fully formalized the proof of Fermat's Last Theorem in Lean, releasing a public repo with over 12 million lines of proof code covering 29k+ theorems.
The author mined the corpus for methods, generalizations, abstractions, and tips for building large Lean proof corpora, packaging them into a new skill.
He also fused in elements from his private math-research skill, designed to instill "epistemic humility": mechanical, structural safeguards that make the agent far more careful about what it asserts and accepts as established, preventing it from jumping the gun. A valuable reference for formal-math and long-chain-reasoning agent engineering.
Related event: Claude Produces First Formalized Proof of Fermat's Last Theorem in 11 Days(52 posts)→
More from coding & agent
- Beat AI's samey frontends: import a design system before generating UI — dotey · 2026-10-21
- Trust slider over Clairic: 102 manual reviews catch 74 of 76 extraction errors — michellechen · 2026-10-21
- FreePi Bets on Ad-Supported Coding Help to Undercut $20/Month AI Coding Subscriptions — heyneighbor · 2026-10-21
- RedMonk: MCP and A2A Verify Publishers, Not Safety — Registries Become the Trust Layer — rseroter · 2026-10-21
- Agentic refactor: 600 PRs moved OpenClaw's web UI from Lit to Solid 2.0 in 2 weeks — steipete · 2026-10-21
- 19 Recorded Runs of 18 Agents Building a Lisp Interpreter, Released on Hugging Face — strickvl · 2026-10-21