What it does This repo collects ready-made LLM prompts for the tedious parts of academic writing: translating Chinese drafts to English LaTeX, stripping “AI taste” from prose, compressing paragraphs to fit page limits, generating figure captions, and even reviewing your own paper from a Reviewer #2 perspective. Everything is copy-paste; no installation, no framework.

The interesting bit The prompts are surprisingly specific about formatting constraints — LaTeX gets escaped percent signs and preserved math mode, Word output must be pure text with zero Markdown, and one prompt demands you “reject \item lists, use coherent paragraphs.” Someone has clearly been burned by LLM slop in camera-ready submissions.

Key highlights

• Prompts sourced from working researchers at MSRA, ByteDance Seed, Shanghai AI Lab, Peking University, USTC, and SJTU
• Separate tracks for LaTeX and Word workflows, plus Chinese-to-Chinese polishing for domestic journal submissions
• Includes “de-AI-ification” prompts trained to remove mechanical connector words and generic phrasing
• New arxiv-translator skill for reading papers: translates LaTeX source and recompiles, not just PDF text extraction
• Agent skills section with setup tutorials for more automated writing pipelines

Caveats

• Entirely in Chinese; English-speaking researchers will need translation or can infer from the markdown code blocks
• No quantitative evaluation of prompt quality — effectiveness claims are anecdotal (“battle-tested by frontline researchers”)
• The “agent skills” section is described as having a learning curve; details are sparse in the truncated README

Verdict Grad students and junior researchers in Chinese-speaking labs will save hours. If you already have a personal prompt library or write in non-CS fields, this is mostly reference material.