Paper Audit

This is a reviewer simulator for academic papers that gives you the kind of structured, blunt feedback you'd get from peer review before you actually submit. It works across LaTeX, Typst, and PDF, supports both English and Chinese, and runs a 16-category issue taxonomy covering everything from formula errors to overclaims to paragraph-level argument incoherence. The deep-review mode dispatches five committee agents (editor, theory, literature, methodology, logic) and separates script-backed findings from LLM judgment calls so you know what's deterministic versus interpretive. Version 4.5 added a mechanical presubmission layer that checks for AI-tone signals, citation hygiene, and abstract completeness. Use it when you want a gate check or revision roadmap, not when you need direct source editing or polishing.