What problem does it solve? AI-generated Chinese academic papers often contain fabricated references, missing in-text citations, template-like phrasing, and unverifiable experimental claims. This Skill enforces a gated six-stage workflow (S0-S5) that requires real literature verification via CNKI browser tasks or scholar_search before any cited content is written, and separates Draft Mode from Final Mode so mock data is never mistaken for submission-ready results. ## Core Features & Use Cases - Verified Literature Gating: Chinese references must come from explicit browser-task CNKI retrieval with page-visible metadata or exported GB/T 7714 citations; English references use scholar_search with layered evidence rules for full-text-level claims. - Discipline-Specific Routing: Routes papers to STEM/medicine or law/humanities reference modules covering IMRaD, CONSORT/STROBE/PRISMA reporting norms, legal citation formats, and journal-specific structures. - Evidence-Based Review: Requires an evidence map, paragraph blueprints, and a six-item S4 audit with quoted textual evidence (Markdown decoration counts, skeleton-reuse comparison tables, strong-claim sentence lists) before delivery. - Use Case: A graduate student asks for a complete thesis draft on defect detection. The Skill first builds a real Chinese and English reference pool, generates a single-file Markdown draft with PLANNING DATA tables and figure placeholders, runs the evidence-backed audit, then converts to a Feishu cloud document. ## Quick Start Ask the assistant to write a Chinese academic paper draft on your topic and specify the paper type, discipline, and any school or journal template you must follow.
