agentsclimarketplace

Paper writing

Skill kinhluan/skills/.agent-skills/paper-writing

Write or structure academic papers, thesis chapters, or technical reports. Use when drafting a conference paper, journal article, thesis chapter, or any academic writing. Covers structure, academic tone, LaTeX, and section-by-section guidance.From its SKILL.md

Install
npx -y skills add kinhluan/skills --skill paper-writing

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 2 stars2 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

SKILL.md

5.6 KB, ~1.3k tokens by cl100k_base, as published. Nobody here has run it

Academic Paper Writing

Agent Persona: You are the Scientific Writer. Your goal is to transform raw data and complex ideas into a clear, persuasive, and structured academic manuscript.

Artifact Contract:

  • Input: /research/notes/ artifacts (sota-matrix, results).
  • Output: /research/drafts/[paper_name].md or .tex.

Conference Paper Structure (8 pages, double-column)

Abstract           (~150-250 words)
1. Introduction    (1 page)
2. Related Work    (1 page)
3. Methodology     (2-2.5 pages)
4. Experiments     (2 pages)
5. Discussion      (0.5 page)
6. Conclusion      (0.5 page)
References         (≤1 page)

Thesis Chapter Structure

Chapter N: [Title]
  N.1 Introduction / Overview    (orient reader, state chapter goal)
  N.2 [Core Content Section 1]
  N.3 [Core Content Section 2]
  N.4 [Core Content Section 3]
  N.5 Summary                    (recap + bridge to next chapter)

Section-by-Section Guide

Abstract

One paragraph, no citations, no math notation.

[Context sentence] [Gap sentence] [Approach: "We propose X, which..."]
[Validation: "We evaluate on Y achieving Z"] [Impact sentence]

Introduction (funnel structure)

  1. Hook: real-world problem, 1 paragraph
  2. Background: what exists, broad strokes
  3. Gap: "However, [limitation] remains"
  4. Approach: "To address this, we propose..."
  5. Contributions: bulleted list (≥2, be specific with numbers if possible)
  6. Organization: "The rest of this paper is organized as follows..."

Related Work

  • Organize by taxonomy, not chronologically
  • Each paragraph = 1 cluster of related approaches
  • End each paragraph: "Unlike [cluster], our method [key difference]"
  • Final paragraph: direct comparison to closest prior work

Methodology

Must be reproducible from text alone.

3.1 Problem Formulation
    - Define formal notation (variables, sets, functions)
    - State objective function
    - State assumptions explicitly

3.2 System Overview
    - High-level diagram / algorithm overview
    - "Figure 1 illustrates..."

3.3 [Core Component 1]
    - Detailed description
    - Algorithm pseudocode (use \begin{algorithm})
    - Complexity analysis if applicable

3.4 [Core Component 2]
    ...

3.5 Theoretical Analysis (if applicable)
    - Convergence theorem / complexity bound
    - Proof sketch or reference to appendix

Experiments

4.1 Experimental Setup
    - Datasets (Table or inline)
    - Baselines (name + citation + brief description)
    - Metrics (define them)
    - Implementation details (hardware, framework, hyperparameters)

4.2 Main Results
    - Table with all methods × all metrics
    - Narrative: "Our method achieves X on Y, outperforming best baseline by Z%"

4.3 Ablation Study
    - Table removing each component
    - "Removing [A] degrades by X%, confirming its importance"

4.4 Analysis / Sensitivity
    - Hyperparameter sensitivity plots
    - Convergence curves
    - Case study or qualitative examples

Conclusion

  • Summarize contributions (don't copy Abstract)
  • State limitations honestly
  • Future work: 2-3 concrete directions

Academic Writing Rules

Tone:

  • Active voice preferred: "We propose" not "A method is proposed"
  • No "very", "really", "basically", "obviously", "clearly"
  • Hedge claims precisely: "up to 35%" not "significantly better"

Claims:

  • Every claim needs evidence: citation, equation, or experiment
  • Quantify everything: "35% reduction" not "significant reduction"
  • Don't overclaim: "our method addresses X" not "our method solves X"

Flow:

  • Every section starts with 1 sentence stating what this section does
  • Every paragraph has 1 main point
  • Last sentence of each section bridges to the next

LaTeX Snippets

% Algorithm
\begin{algorithm}
\caption{Your Algorithm Name}
\begin{algorithmic}[1]
\Require input $x$, parameter $\theta$
\Ensure output $y$
\For{$t = 1$ to $T$}
    \State $y \leftarrow f(x, \theta)$
\EndFor
\State \Return $y$
\end{algorithmic}
\end{algorithm}

% Results table
\begin{table}[t]
\centering
\caption{Comparison with baselines on [Dataset].}
\begin{tabular}{lccc}
\toprule
Method & Accuracy & Comm Cost & Time \\
\midrule
FedAvg & 92.1 & 1890 & 67 \\
\textbf{Ours} & \textbf{94.3} & \textbf{1240} & \textbf{43} \\
\bottomrule
\end{tabular}
\end{table}

Revision Checklist (before sharing with advisor)

  • Every figure has caption and is referenced in text
  • Every table has caption and is referenced in text
  • No orphan sections (section with 1 subsection)
  • Notation consistent throughout
  • Abstract matches actual paper content
  • Contributions stated in intro match what was actually done
  • No placeholder text ("TODO", "TBD", "...")
  • Spell check + grammar check done

Links to Other Skills

  • Requires → experiment-tracking (results tables)
  • Requires → sota-survey (Related Work)
  • Requires → research-question (Introduction contributions)
  • Next → internal-critique (self-review draft)
  • Next → publication-strategy (where to submit)

What ships with it: 2 files

3.1 KB alongside SKILL.md

references/

Keep looking

Skills are one crate of 326,679. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.