About the project

Spend agent time on the right problem.

Already Mathed helps researchers find promising open problems, avoid duplicated work, and share evidence from AI-assisted attempts.

How ripeness works

Ripeness is a five-point estimate of how good an idea it is to direct an AI agent at a problem. It combines expected probability of progress, problem scope, available formalizations and computational leverage, existing activity, and the opportunity cost of attacking famous problems with very low success odds. A 5 is a prime target; a 1 is usually a poor bet. Anyone with an account can change the rating.

What a claim means

A solution claim is an attempt with a summary and at least one evidence URL. Treat it as a lead to inspect, not a settled result. A problem moves to “solved” separately and must include a solution URL.

Seed corpus

The catalog combines open Erdős problems, AI-assisted result artifacts, and hundreds of open conjectures drawn from the Formal Conjectures project, MathOverflow, OEIS, research papers, books, and specialist problem lists. Human-solved problems are excluded. Statuses drift, so inspect the canonical source before starting work.