[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-researchers-evolve-a-cheaper-interface-for-ai-theorem-provers":10,"sections":40},{"siteName":4,"siteTagline":5,"publisherName":4,"contactEmail":6},"The Revision","Tech news, decoded.","editor@therevision.news",{"gaMeasurementId":8,"adsenseClientId":9},"G-ZW2MV82GYR","ca-pub-8533917693782264",{"article":11},{"id":12,"slug":13,"title":14,"dek":15,"body_md":16,"tags_json":17,"published_at":18,"created_at":19,"updated_at":20,"status":21,"review_note":22,"review_notes":23,"image_url":22,"persona_id":22,"persona_name":22,"section":30,"tags":31,"sources":35,"feedback":39,"feedback_at":22,"cost_usd":39,"total_tokens":39},8947,"researchers-evolve-a-cheaper-interface-for-ai-theorem-provers","Researchers Evolve a Cheaper Interface for AI Theorem Provers","A new evolved MCP server lets AI agents prove math theorems in Rocq and Lean faster and cheaper than existing tools, without human redesign of the interface.","AI agents that prove math theorems now have a cheaper way to talk to the software that checks their work.\n\nResearchers built a new interface for connecting AI agents to Rocq and Lean, two proof assistants that verify machine-generated math proofs step by step. Instead of hand-designing that interface the way humans always have, they let a large frontier model propose new features for it, then tested each change against smaller models and kept only the ones that actually helped. The result is a tool the paper calls RME, built as a Model Context Protocol (MCP) server, a standard format for letting AI agents call external tools, for the Rocq prover. On the held-out test set of miniF2F-Rocq, agents using RME beat both a bare-bones setup exposing just the Rocq compiler and an existing MCP server, posting higher success rates while spending less money and time per solved problem, across four models from two different model families. The gains carried over to Lean too, improving cost and time per solve on a subset of PutnamBench.\n\nThis matters because the interface between AI and proof assistants has mostly been borrowed from tools built for human mathematicians, not tuned for how automated agents actually work, and that mismatch burns money and time on every attempt. Letting a model design and prune its own tool is a concrete example of AI improving the scaffolding it runs on, not just the reasoning inside it. That the gains transfer from Rocq to a different proof assistant in Lean suggests the method found something general about what agents need, not a quirk of one tool.\n\nIt is a narrow result, tested on a few hundred competition-style problems, but it is a reminder that a lot of current AI math progress is about plumbing, not just bigger models.","[\"ai\",\"ai-agents\",\"theorem-proving\",\"research\"]","2026-10-01T04:00:00.000Z","2026-10-01T12:14:52.550Z","2026-10-01T12:14:57.024Z","published",null,[24],{"id":25,"reviewer":26,"round":27,"reason":28,"status":29},"editor-r1","editor",1,"Define the acronym MCP (Model Context Protocol) on its first use in the body — it's currently used without ever being spelled out.","resolved","ai",[30,32,33,34],"ai-agents","theorem-proving","research",[36],{"name":37,"url":38},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.39544",0,{"sections":41},[42,45,50,55,60,65,70,75,80,84,89,94,99,104],{"name":43,"slug":30,"count":44,"latest_published_at":18},"AI",5350,{"name":46,"slug":47,"count":48,"latest_published_at":49},"Security","security",801,"2026-09-30T22:18:23.000Z",{"name":51,"slug":52,"count":53,"latest_published_at":54},"Policy","policy",429,"2026-10-01T02:26:17.000Z",{"name":56,"slug":57,"count":58,"latest_published_at":59},"Deals","deals",298,"2026-09-30T21:00:26.000Z",{"name":61,"slug":62,"count":63,"latest_published_at":64},"Hardware","hardware",196,"2026-09-30T13:00:00.000Z",{"name":66,"slug":67,"count":68,"latest_published_at":69},"Science","science",157,"2026-09-30T15:00:56.000Z",{"name":71,"slug":72,"count":73,"latest_published_at":74},"Consumer Tech","consumer-tech",149,"2026-09-30T22:57:11.000Z",{"name":76,"slug":77,"count":78,"latest_published_at":79},"Dev Tools","dev-tools",93,"2026-10-01T02:30:48.000Z",{"name":81,"slug":82,"count":78,"latest_published_at":83},"Software","software","2026-09-30T21:41:11.000Z",{"name":85,"slug":86,"count":87,"latest_published_at":88},"Startups","startups",84,"2026-09-30T20:39:09.000Z",{"name":90,"slug":91,"count":92,"latest_published_at":93},"Gaming","gaming",51,"2026-09-30T16:24:30.000Z",{"name":95,"slug":96,"count":97,"latest_published_at":98},"General","general",50,"2026-09-30T21:37:54.000Z",{"name":100,"slug":101,"count":102,"latest_published_at":103},"Reviews","reviews",31,"2026-09-28T14:31:34.000Z",{"name":105,"slug":106,"count":107,"latest_published_at":108},"How-To","how-to",6,"2026-06-16T09:00:00.000Z"]