What problem does it solve? Writing academic papers for venues like Speech Communication, IEEE/ACM Transactions, Interspeech, and ICASSP involves high risks of AI-generated hallucinations (fabricated references, invented statistics), inconsistent English style, and LaTeX compilation errors. This Skill provides an end-to-end workflow that enforces source-traceable numbers, verified citations, and field-appropriate writing style. ## Core Features & Use Cases - Hallucination Prevention: Enforces a three-principle citation rule (well-known, PDF-downloadable, content-verified) and classifies every numeric value as directly sourced, derived, Python-verified, or unverified, with unverified values strictly forbidden. - End-to-End Workflow: Covers project setup, source scanning, related-work table verification, section-by-section drafting, numeric validation scripts, compilation via compile.sh, and submission packaging. - Style Control: Ships corpus-measured style data (62,977 papers from ICASSP/Interspeech) defining banned AI-sounding words, human-marker phrases, sentence-length targets, and venue-specific conventions. - Use Case: When drafting a journal paper from a master's thesis and prior conference papers, the Skill extracts all facts from the reference folder, builds a verified bibliography spreadsheet, writes each section in recommended order, and produces a submission-ready PDF with revision marks toggled off. ## Quick Start Ask the AI to start a new paper writing project using the paper_writing skill and point it to your reference materials folder.
