fakerqwq/social-science-paper-writing-skill

This repository provides an AI agent skill designed to assist students and researchers writing social science papers. It is built for use with AI coding assistants that support skills or can load instruction files as context. The goal described in the README is not to polish prose but to help writers produce papers that are verifiable, properly argued, and where the evidence and methods actually match the claims being made. The skill covers nine distinct stages of the paper writing process. Topic diagnosis checks whether a proposed research topic has genuine scholarly value, feasibility, and an evidence base. Research question refinement takes a broad theme and helps narrow it to a specific, answerable question. Outline building checks that the title, research question, literature review, method, analysis, and conclusion are consistent with one another. Literature review planning helps organize sources by debate, concept, variable, mechanism, method, or case type rather than listing authors in sequence. A literature search workflow supports moving from database searches in CNKI or Google Scholar through to Zotero for citation management. Full draft review diagnoses an entire paper across argumentation, structure, theory, method, evidence, and citations. Section revision rewrites individual chapters or paragraphs without inventing evidence. Citation and evidence checking flags four specific risk types: missing evidence for a claim, unverifiable or mismatched citations, direct quotations without page numbers, and causal language that goes beyond what the research design can support. Pre-submission checking gives a final pass-or-revise judgment before the paper is submitted. A firm rule stated throughout the README is that the skill will not fabricate titles, data, citations, page numbers, or conclusions. It distinguishes between sources that have only been found as metadata, sources where only the abstract was read, sources where the full text was read, and sources that have been imported to Zotero. The repository structure includes the main skill file, a checklist, citation rules, methodology rubrics, paper structure guides, a revision workflow, example outlines, and reference documents for literature review and research design. Installation involves cloning the repository into a local skills directory and then invoking it by name in the AI assistant. The README is written entirely in Chinese.