What problem does it solve? Mathematical proofs in theory papers often contain hidden gaps, unjustified interchanges, and unstated assumptions that are hard to catch through self-review. This Skill systematically audits a LaTeX proof using an independent cross-model reviewer, fixes each identified gap with full derivations, and produces a formal audit report. ## Core Features & Use Cases - Cross-Model Adversarial Review: Submits the full proof to a non-Claude reviewer backend (Codex or Manual Review MCP) at ultra reasoning effort, checking hypothesis discharge, inequality chains, limit interchanges, quantifier order, and degenerate cases against a 20-category issue taxonomy. - Proof-Obligation Ledger: Builds a dependency DAG, assumption ledger, typed symbol table, canonical quantified statements, and micro-claim inventory in PROOF_SKELETON.md before judging validity. - Iterative Fix-and-Re-Review Loop: Applies severity-ordered fixes (FATAL to MINOR) directly in the .tex source, re-reviews up to 3 rounds, runs blind second reviews on critical fixes, and compiles with pdflatex to confirm no errors. - Use Case: A researcher finishing a theory paper runs the skill on main.tex; it flags an illegal dominated-convergence interchange in Lemma 4, derives the missing domination argument, patches the LaTeX, and emits PROOF_AUDIT.md plus a formal proof_audit_report.tex. ## Quick Start Ask the agent to verify the proof in your paper's main.tex file, optionally enabling --deep-fix for repair-grade fix plans or --restatement-check to detect theorem statement drift across sections.
