SkillAgentSearch skills...

paper-polish

Proofread and verify academic papers in LaTeX. Runs 19 sequential checks covering titles, consistency, citations, formatting, theoretical tension in motivation, concise results reporting, cross-section repetition, em-dash usage, and auxiliary-text-to-footnote conversion.

Install / Use

npx skills add brycewang-stanford/Auto-Empirical-Research-Skills --skill paper-polish

Installs into whichever agent you are using.

About this skill
📄

SKILL.md

Installable skill definition

Quality Score

95/100

Supported Platforms

Universal

Tags

Our assessment of paper-polish

paper-polish scores 95/100 on our quality scale, 48th of 421 Education & Research skills we index (top 12%).

Its SKILL.md is 28 KB long, well organised into 42 sections with 19 code examples: a thorough specification that gives an agent plenty to work with.

With 4,360 GitHub stars, it is one of the more widely adopted skills in the catalogue.

Substance
30/30
Structure
20/20
Description
15/15
Adoption
15/20
Freshness
15/15

Maintenance, license and trust

  • The repository was last updated about 2 months ago, so paper-polish is actively maintained.
  • Our last check on 2026-11-15 found the source still online.
  • No license is declared. By default that means all rights are reserved: you can read it, but reusing or redistributing it is not clearly permitted. Ask the author before building on it commercially.
  • Its trust signals score 88/100, with 1 caution from licensing, adoption, age or documentation. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.

Safety scan

No issues found

Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful.

AI review by kimi-k2.7-code on 2026-09-29. Automated pattern scan on 2026-09-29. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.

paper-polish compared with similar skills

All 4 of these similar skills score higher than paper-polish; compare them before choosing.

SkillScoreStarsUpdatedFormat
paper-polish (this skill)by brycewang-stanford954.4k54d agoSKILL.md
last30days-skillby mvanhorn10068.0ktodayCLAUDE.md
x402by browser-use100121.0k3d agoSKILL.md
algorithmic-artby anthropics100177.9k55d agoSKILL.md
pptxby anthropics100177.9k55d agoSKILL.md

Frequently asked questions

How do I install paper-polish?
Run npx skills add brycewang-stanford/Auto-Empirical-Research-Skills --skill paper-polish. The install tabs above show the steps for each supported agent.
Which AI agents does paper-polish work with?
It is written for Universal, as a SKILL.md file. Other agents that read the same format can often use it too.
Is paper-polish safe to use?
Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful. It declares no license and scores 88/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
Is paper-polish still maintained?
The repository was last updated about 2 months ago, so paper-polish is actively maintained.

name: paper-polish description: Proofread and verify academic papers in LaTeX. Runs 19 sequential checks covering titles, consistency, citations, formatting, theoretical tension in motivation, concise results reporting, cross-section repetition, em-dash usage, and auxiliary-text-to-footnote conversion. Trigger when user says "check paper" / "proofread" / "paper-checker" / "校对" / "核查论文". allowed-tools: Read, Bash, Edit, Write, Glob, Grep, AskUserQuestion, WebSearch, WebFetch argument-hint: "[path-to-paper-directory]"

Paper Checker

Overview

This skill performs a comprehensive 19-step proofreading and verification of academic papers in LaTeX. Each step is run sequentially. Within each step, every individual issue found is presented to the user one at a time for approval before applying the change.

CRITICAL INTERACTION RULE: For every issue found in every check, you MUST:

  1. Present the issue clearly (original vs. proposed change, or finding).
  2. Ask the user: "Accept this change? (yes/no/edit)"
  3. Wait for the user's response before proceeding to the next issue.
  4. If the user says "yes", apply the change using the Edit tool.
  5. If the user says "no", skip and move to the next issue.
  6. If the user says "edit" or provides alternative text, apply the user's version instead.
  7. After all issues in a check are resolved, announce completion and move to the next check.

All output must be in English unless the user specifies otherwise.

Initialization

  1. Receive the paper directory path from user (via $ARGUMENTS or ask).
  2. Scan the directory to identify:
    • Main LaTeX file (main.tex)
    • Bibliography file (ref.bib or *.bib)
    • Results directory with table/figure .tex files
    • Any PDF with original tables/figures
  3. Read all files into context (main.tex in chunks if large, all table .tex files, ref.bib, PDF pages).
  4. Present inventory and confirm with user before proceeding.

The 19 Checks

Run these checks in order. For each check, present ALL findings, then ask the user to approve/reject each one individually.


Check 1: Section and Subsection Titles

Optimize all \section{} and \subsection{} titles to reflect full academic rigor. Modified titles must NOT use colons.

Output format (present one row at a time for approval):

## Check 1: Section & Subsection Titles

Issue 1/N:
| Original Title | Proposed Title |
|----------------|----------------|
| Robustness Checks | Sensitivity and Robustness Analyses |

Accept this change? (yes/no/edit)

Check 2: Table and Figure Titles

Optimize all \caption{} in table/figure .tex files to reflect full academic rigor. Modified titles must NOT use colons.

Output format (one at a time):

## Check 2: Table & Figure Titles

Issue 1/N:
| Original Title | Proposed Title |
|----------------|----------------|
| Summary Statistics | Descriptive Statistics of Key Variables |

Accept this change? (yes/no/edit)

Check 3: Internal Consistency of Arguments

Check the entire paper for contradictory or inconsistent claims across sections.

Output format (one issue at a time):

## Check 3: Internal Consistency

Issue 1/N:
Location: [Section X, paragraph Y] vs. [Section Z, paragraph W]
Problem: [description of inconsistency]

| Original Text | Proposed Revision |
|---------------|-------------------|
| "...excerpt..." | "...**revised** excerpt..." |

Accept this change? (yes/no/edit)

Check 4: Introduction Logic and Flow

Check whether the introduction's logic is tight and flows smoothly from motivation to findings to contributions.

Output format (one issue at a time):

## Check 4: Introduction Logic

Issue 1/N:
Location: [paragraph N of Introduction]
Problem: [e.g., abrupt transition, missing logical link, redundancy]

| Original Paragraph | Proposed Revision |
|---------------------|-------------------|
| "...excerpt..." | "...**revised** excerpt..." |

Accept this change? (yes/no/edit)

Check 5: Storyline Coherence

Evaluate whether the full paper's storyline is tight and coherent. If any table is not essential to the main story, suggest moving it to the appendix.

Output:

  • Assessment of overall storyline flow
  • For each potentially non-essential table:
Table X: [name]
Recommendation: Move to appendix / Keep in main body
Rationale: [explanation]

Accept this recommendation? (yes/no)

Check 6: Theoretical Tension in Introduction Motivation

Evaluate whether the motivation paragraphs of the introduction establish sufficient theoretical tension — i.e., competing economic forces or theoretical perspectives that make the central research question genuinely interesting ex ante. A motivation that points in only one direction reads as "the answer is obvious, so why run this regression?" and weakens the paper's contribution.

What to look for:

  • The central $X \to Y$ relationship the paper examines is identified, but only one mechanism / one direction is articulated (e.g., the intro only argues "$X$ raises $Y$" with no countervailing reason $X$ could lower $Y$ or leave $Y$ unchanged).
  • Prior literature is described as a one-directional consensus instead of as a genuine debate or open question.
  • The motivation paragraphs read like a summary of the eventual finding ("we show that $X$ raises $Y$, which is consistent with [single theory]") rather than as a setup of a puzzle.
  • No mention of plausible alternative predictions, opposing theoretical traditions, or institutional features that could push the effect the other way.
  • The contribution is framed as "we confirm [obvious prior]" rather than "we resolve a tension between [view A] and [view B]".
  • Hypotheses development (if previewed in intro) immediately commits to a signed prediction without acknowledging that the opposite is also plausible under a different framework.

How to revise — apply these rules:

  1. Identify the focal $X \to Y$ relationship and the paper's actual finding (do NOT change the finding).
  2. Articulate at least two competing predictions that are each individually plausible before the data is examined. Typical sources of tension:
    • Opposing theoretical traditions (e.g., agency theory vs. stakeholder theory; risk-shifting vs. risk-management; information asymmetry vs. monitoring).
    • Direct vs. indirect / countervailing channels (e.g., $X$ raises $Y$ through channel A but lowers $Y$ through channel B).
    • Conflicting prior empirical findings in the literature.
    • Institutional or contextual features that could flip the sign (e.g., regulation, ownership structure, market frictions).
  3. Reframe the motivation as a puzzle, with a structure such as:

    "On the one hand, [theory / mechanism A] predicts that $X$ should increase $Y$, because [reason]. On the other hand, [theory / mechanism B] predicts that $X$ should decrease $Y$ (or leave it unchanged), because [reason]. Which force dominates is ultimately an empirical question."

  4. Cite competing prior work to anchor each side of the tension. If one side has no direct prior, cite the theoretical paper or a related literature that motivates that prediction.
  5. Position the contribution as resolving the tension, not as confirming an expected result. Phrasing like "we provide evidence that helps adjudicate between …" or "our findings reconcile …" is stronger than "we document that …".
  6. Do not introduce strawman alternatives. The competing prediction must be one a reasonable referee in the field would actually find plausible.
  7. Do not contradict the paper's actual findings. Tension is about the ex ante prediction, not the ex post result.
  8. Do not invent mechanisms not supported by economics / finance / accounting literature. If unsure, search for relevant theoretical or empirical citations before proposing the alternative force.

Severity assessment — flag the motivation as deficient in theoretical tension if any of the following hold:

  • Only one signed prediction is articulated in the motivation paragraphs.
  • No prior work is cited as offering a competing view.
  • The motivation paragraph and the hypothesis statement give the same one-directional argument with no opposing consideration.
  • The wording reveals the empirical sign before the puzzle is established (e.g., "We argue and find that $X$ raises $Y$" stated up front, with no setup of why the opposite was plausible).

Output format (one issue at a time):

## Check 6: Theoretical Tension in Introduction Motivation

Issue 1/N:
Location: Introduction, paragraph(s) [N]
Focal relationship: $X \to Y$ ([brief description of X and Y])
Problem: [e.g., "Only one mechanism is articulated; no countervailing force is acknowledged; prior literature framed as one-directional consensus."]
Proposed competing prediction: [one-sentence summary of the alternative force / theory and its source]

| Original Motivation Excerpt | Proposed Revision |
|------------------------------|-------------------|
| "...[single-directional motivation passage]..." | "**On the one hand, [mechanism A] suggests that $X$ raises $Y$, because [reason A] (Author, Year). On the other hand, [mechanism B] suggests that $X$ may instead lower $Y$, because [reason B] (Author, Year). Which force dominates is an empirical question.**" |

Accept this change? (yes/no/edit)

If the motivation already exhibits adequate theoretical tension (at least one explicitly stated, plausibly opposing prediction with a citation), report:

## Check 6: Theoretical Tension in Introduction Motivation

Assessment: The introduction's motivation establishes adequate theoretical tension between [mechanism A] and [mechanism B], with prior literature cited on both sides. No revision proposed.

Check 7: Section Ordering

Check whether the ordering of sections and subsections is logical and follows academic conventions for the target field.

Output: Assessment of current ordering with specific reordering suggestions if needed. Present each suggestion for approval.


Check 8: Table/Figure References in Text

Verify that every table and figure included in the paper is mentioned/referenced in the main text.

Output format:

## Check 8: Unreferenced Tables/Figures

| # | Table/Figure | File | Referenced in Text? |
|---|-------------|------|---------------------|
| 1 | Table X | results/xxx.tex | NOT referenced |

[If all referenced]: All tables and figures are referenced in the text.

For unreferenced items, suggest where to add a reference. Ask user for approval.


Check 9: Variable Definition Table Completeness

Check whether the variable definition table (if present) includes all key variables discussed in the text.

Output format:

## Check 9: Missing Variables in Definition Table

| # | Variable | Mentioned in Section | Missing from Definition Table |
|---|----------|---------------------|-------------------------------|
| 1 | FiscalTransparency | Section 6.3 | Yes |

Suggest adding these variables? (yes/no)

Check 10: Abbreviation and Nomenclature Consistency

Check that all abbreviations/acronyms are defined on first use and used consistently throughout.

Output format (one issue at a time):

## Check 10: Abbreviation Issues

Issue 1/N:
| Abbreviation | Issue | Location | Suggested Fix |
|-------------|-------|----------|---------------|
| PSM | Used without definition | Section 6.2 | Define as "propensity score matching (PSM)" on first use |

Accept this fix? (yes/no/edit)

Check 11: Content and Logic Completeness

Check whether the paper's arguments, reasoning, and evidence are complete. Flag any gaps where claims lack support or transitions are missing.

Output format (one issue at a time):

## Check 11: Content/Logic Gaps

Issue 1/N:
Location: [Section, paragraph]
Gap: [description]

| Original 

Truncated for display — read the full file on GitHub.

Related Skills

View on GitHub
GitHub Stars4.4k
CategoryEducation
Updated1mo ago
Forks527

Languages

Stata

Trust signals

88/100

From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.

1 medium