What problem does it solve? Competition math proofs (IMO, Putnam, USAMO, AIME) often contain subtle errors that self-verification misses, and AI solvers frequently bluff confident but wrong answers. This Skill orchestrates parallel solving agents with fresh-context adversarial verification to produce rigorously checked proofs or honest abstentions. ## Core Features & Use Cases - Parallel candidate generation: Launches 8-12 solver agents with varied starting angles, each internally iterating solve, self-improve, and self-verify cycles. - Adversarial verification: Fresh-context verifiers attack cleaned proofs using 12 specific failure patterns (tautological reductions, proves-too-much, one-line lemmas) with asymmetric 4-confirm/2-refute voting and early exit. - Calibrated abstention and deep mode: Says "no confident solution" rather than guessing, with an optional deep-mode agent allowing bounded local computation before abstaining. - Use Case: Given an IMO problem statement, the Skill generates multiple candidate proofs, strips reasoning traces, runs adversarial verification passes, revises holes up to 3 cycles, and compiles the verified proof into a LaTeX PDF. ## Quick Start Ask the assistant to solve this IMO problem and verify the proof adversarially before presenting it as a PDF.
