AI Agent Hub
Back to skills
⚖️

Research Paper Writing Assistant

Professional Updated 2026.08.30

Paste the following prompt into your AI chat to install this skill:

Please follow https://skillhub.cn/install/skillhub.md and install @org-02qudk26/research-paper-writing.

About this skill

What It Solves

Research papers often sound plausible while the argument remains weak: claims in the Abstract or Introduction lack experimental support, key terms depend on hidden context, and figures, tables, and layout reduce first-impression quality. This skill treats rewriting as reviewer-side structured revision, not just sentence polishing, and focuses on narrative line, paragraph flow, terminology consistency, and claim-evidence alignment.

How It Works

It starts by clarifying the paper story, then loads only the relevant section guide from references/, such as introduction.md, method.md, or experiments.md. Writing follows a one paragraph, one message rule: the first sentence states the paragraph role, new terms are defined before reuse, and sentences are connected by cause, contrast, consequence, refinement, or example. After each section, it runs reverse outlining to check whether the thesis, topic sentences, and evidence map cleanly; paragraphs that cannot be mapped are revised or removed. A final adversarial review uses paper-review.md and covers contribution, clarity, experimental strength, evaluation completeness, and method design.

Where It Fits

It is useful when experimental materials already exist and the goal is to rewrite sections or prepare a reviewer-friendly draft, especially for the Abstract, Introduction, and major claims. The output includes a compact outline, paragraphs with explicit roles, a self-review checklist, and a Claim | Evidence | Status map. It does not replace experimental design or statistical validation; unsupported claims should be weakened or removed according to its rules.

Use Cases

  • Map claims to evidence and delete unsupported claims.
  • Rewrite Method paragraphs with topic first and clear links.
  • Run five-dimension self-review for contribution and evidence.
  • Condense Related Work into motivation, contrast, and gap.

Best For

  • PhD student: rewrite abstract and map claims to results.
  • ML engineer: check Method paragraphs for design and gap.
  • Co-author: run reverse outlining and reviewer questions.
  • Conference author: create claim-evidence map and checklist.