references/academic-phrasebank.md
<!-- Canonical copy. If this file is duplicated into another skill, edit THIS copy
(paper-polish/references/) and re-sync; scripts/check_shared_sync.py enforces parity. -->
# Academic phrasebank
Alternatives organized by purpose, for when finer wording support is needed. Pick
the plainest phrase that fits; do not stack connectives into every sentence.
Chinese equivalents are kept because this phrasebank also serves
Chinese-to-English rewriting.
## Verbs by evidence strength
Matching the verb to the strength of the evidence is the most practical calibration
habit.
**Strong (solid design and data)**
- English: show, demonstrate, establish, reveal, identify
- Chinese: 表明、证实、确立、揭示
**Medium (reasonable but not certain)**
- English: suggest, indicate, support the view that, be consistent with, point to
- Chinese: 提示、说明、支持……的看法、与……一致、指向
**Weak (beyond direct observation, speculative)**
- English: may reflect, could arise from, appears to, seems likely, might be
explained by
- Chinese: 可能反映、或源于、似乎、看来、或可解释为
Describing the evidence itself:
- weak: limited, scant, insufficient
- accumulating: growing, emerging, accumulating
- strong: robust, reliable, convincing, considerable
## Describing the gap (precise, not dramatic)
A good gap statement is exact, not theatrical.
- Use: remains poorly understood / has not been examined in... / has received
limited attention / few studies have addressed... / evidence remains sparse for...
- Chinese: 仍不甚清楚 / 尚未在……情形下被检验 / 受到的关注有限 / 鲜有研究处理…… /
关于……的证据仍很稀少
- Avoid: no one has ever studied / completely unknown / ignored by all previous
work. These almost never survive review.
## Comparing with prior work
Alignment (this work agrees with prior reports):
- These results are consistent with... / This finding accords with... / Our
observations broadly support...
- Chinese: 这些结果与……一致 / 这一发现与……相符 / 我们的观察大体支持……
Marking the difference (fairly, without belittling):
- In contrast to earlier reports... / This finding differs from... / One possible
reason for this discrepancy is...
- Chinese: 与早先的报道不同…… / 这一发现不同于…… / 造成这一差异的一个可能原因是……
Stating the gap fairly instead of building a straw man:
- Although previous studies showed..., their performance in... remains unclear.
- Earlier work established..., but did not address...
- Chinese: 尽管已有研究展示了……，但它们在……方面的表现仍不清楚。 /
早先的工作确立了……，但没有处理……。
## Limitations
- These findings should be interpreted with caution because... / A limitation of
this study is that... / The generalisability of these results is limited by... /
We cannot exclude the possibility that...
- Chinese: 这些发现应谨慎解读，因为…… / 本研究的一个局限是…… /
这些结果的可推广性受限于…… / 我们无法排除……的可能。
Tie each limitation to a real source of uncertainty; never write it as an empty
courtesy.
## Implications and future work
Implications (do not step past the evidence):
- An implication of this is that... / These findings may help to explain... /
This work has implications for...
- Chinese: 由此引出的一个含义是…… / 这些发现或有助于解释…… / 这项工作对……有启示。
Future work (should grow out of a real limitation or opportunity):
- Further work is needed to determine whether... / Future studies should
examine... / A useful next step would be to...
- Chinese: 还需进一步工作来确定是否…… / 未来研究应考察…… / 一个有用的下一步是……
## Transitions (use the smallest one that works)
- Contrast: however, by contrast, nevertheless, despite this, whereas
- Addition: furthermore, in addition, moreover, also
- Cause: therefore, thus, consequently, as a result, thereby
- Qualification: notably, importantly, approximately, in part, at least in this
cohort
Do not open consecutive paragraphs with "This suggests...". Alternatives: repeat
the noun (Such heterogeneity...), use a definite noun phrase (The resulting
gradient...), use a participial summary (Taken together, ...), or, when the logic
is already clear, connect with nothing at all. One demonstrative-opening per
paragraph is enough.
references/ai-tone-guardrails.md
<!-- Canonical copy. If this file is duplicated into another skill, edit THIS copy
(paper-polish/references/) and re-sync; scripts/check_shared_sync.py enforces parity. -->
# AI-tone guardrails and inflated-claim wordlist
Scan against these two lists when removing AI flavor or calibrating claim strength.
A hit does not automatically mean "delete"; it means stop and ask whether the word
carries information or merely performs depth.
## Typical AI-tone signals
Scan item by item; rewrite on hit:
- **Grandiose framing**: stands as a testament, is a testament to, pivotal moment,
evolving landscape, reflects a broader, setting the stage, marking a shift.
Delete; replace with concrete facts.
- **Shallow -ing tails**: highlighting / underscoring / emphasizing / ensuring /
reflecting / showcasing hung at the end of a sentence to fake depth. Expand into
a real clause or delete.
- **Marketing adjectives**: boasts, vibrant, rich (figurative), profound, nestled,
in the heart of, groundbreaking, breathtaking. Delete.
- **Vague attribution**: industry reports suggest, observers have cited, experts
argue, with no source. Name the real source or delete.
- **High-frequency AI words**: delve, intricate, tapestry, underscore, testament,
garner, pivotal, vibrant, leverage, empower (as marketing verbs), encompass,
stems from, pave the way for, notably, yielding, at its essence, impede.
Replace with plain words.
- **Copula avoidance**: serves as / stands as / represents / boasts. Use is / are / has.
- **Negative parallelism, forced triads, fake ranges**: not only... but also...,
it is not just... it is..., forcing three items, from X to Y when the two ends are
not on one scale. State directly.
- **Synonym cycling**: the same object called three different names in one passage.
Pick one name and keep it.
- **Hyphenated-pair pileup**: third-party, data-driven, end-to-end, real-time,
cutting-edge stacked as decoration. Keep necessary terms, cut ornaments.
- **Fragmented headings**: a heading followed by one line that restates the heading.
Delete or merge into prose.
- **Formatting noise**: excessive dashes, mechanical bolding, Title Case headings,
emoji, curly quotes.
- **Chat politeness**: Great question, I hope this helps, as of my last update,
You are absolutely right. Delete.
- **Filler and over-hedging**: in order to (use to), due to the fact that (use
because), could potentially possibly (use may), empty upbeat closers.
Closing self-check: ask "where does this passage still read machine-written?", list
the remaining signals, then rewrite so it does not.
Reminder: after stripping every signal, prose can turn clean but lifeless, with
uniform sentence length and no acknowledged complexity. The fix, still without
inventing facts or inflating claims: vary sentence length, carry weight with exact
facts instead of stacked adjectives, and state real trade-offs and limitations
plainly. Voice comes from precision and honesty, not from noise.
## Inflated-claim wordlist (soften or flag unless the data genuinely supports it)
Do not auto-generate high-risk claims such as "first", "only", "best". When the
author's own draft contains them, do not silently keep them either: soften, or flag
for the author to confirm the evidence.
**English**: extraordinary, exceptional, remarkable, unprecedented, revolutionary,
groundbreaking, new paradigm, best, superior, state-of-the-art, proves,
conclusively, perfectly, first (without evidence), innovative, pioneering,
transformative, surpass, excel, breakthrough.
**Chinese**: 卓越、首次、唯一、最高、完美、碾压、遥遥领先、颠覆性、最优。
Safer replacement directions:
- prove: show / suggest, depending on evidence
- conclusively: delete, or "to our knowledge"
- best / superior: among the strongest / performs competitively
- unprecedented: delete, or bound to a specific scope
- first: "to our knowledge, the first" (only when actually checked), or bound to
the studied cohort or setting
The point is not to weaken every strong word. It is to make claim strength match
evidence strength. When the evidence is strong and the scope is stated, strong
verbs are correct.
## One-line summary
De-AI-toning and de-inflation are two faces of the same job: make the text honest.
No tone that carries zero information; no claim the evidence cannot carry. Edit the
expression, never the facts: do not invent data, experiments, or citations to prop
up a strong conclusion.
references/section-conventions.md
<!-- Canonical copy. If this file is duplicated into another skill, edit THIS copy
(paper-polish/references/) and re-sync; scripts/check_shared_sync.py enforces parity. -->
# Per-section writing conventions
## Table of contents
1. Introduction
2. Methods
3. Results
4. Discussion
5. Conclusion
6. Abstract
7. Title
Each section has its own job and conventions. Consult this when polishing or
drafting needs finer per-section guidance. Each entry lists the questions the
section must answer, the recommended order, and the common faults.
## Introduction
Must answer: why does this topic matter? What is known? What is missing or
contested? What exactly does this paper ask, and how?
Recommended order: establish importance, summarize what is known, name the gap or
controversy, state the goal, point to the value or the route.
Common faults:
- Opens like a textbook background instead of a positioning move.
- The gap is only implied, never stated.
- The step from "what is known" to "what this paper does" skips "why is this still
needed".
- Spoils the method or the results in detail.
Avoid: long historical build-up, detailed results, and inflated novelty claims
before the gap is defined.
## Methods
Must answer: could another team reproduce the work from this description alone,
or from this description plus one explicitly cited protocol?
Recommended order: design or cohort, materials or data sources, procedure, outcome
measures, analysis and statistics, ethics statement when applicable.
Watch for empty phrases that say nothing; replace them with reproducible facts:
- under standard conditions
- using routine methods
- data were analysed statistically
- differences were significant (no test, no effect size)
- the method was validated (without saying how)
Rely on a citation to omit detail only when the cited report actually contains the
needed detail.
## Results
Must answer: what was observed, under what conditions, with what quantitative
support?
Recommended order: point the reader to the figure, table, or experiment; state the
main observation; add quantitative detail; note expected or unexpected patterns;
compare with prior work only when it clarifies the result.
Use past tense. Typical verbs: was detected, increased, showed, achieved.
Common faults:
- Inline interpretation inside Results (that is the Discussion's job): limit
may reflect, suggests that, is likely due to, unless used as a deliberate bridge.
- Results that belong in the main text pushed into supplementary material.
- Vague comparisons (higher than control) without effect size or test.
## Discussion
Must answer: what do the main findings mean? How do they relate to prior work?
Which interpretations are plausible? Which limitations constrain the reading? What
follows, and what does not?
Recommended order: restate the main finding, give plausible interpretations,
compare with prior work, name limitations, state implications, point to future
work when useful.
This is where hedging belongs.
Common faults: rewriting the Results section in different words; describing a
correlation as a mechanism.
## Conclusion
Must answer: what is the core contribution? Which finding is decisive? What
implication follows, and within what boundary?
Recommended order: return to the goal, summarize the decisive finding, state the
contribution or significance, give the boundary or outlook.
The conclusion is not an abbreviated Discussion. Do not introduce new experiments
here, and do not close with generic praise of the work.
## Abstract
Must answer: what problem or gap? What was done? What was found? Why should the
reader care?
Recommended order: background, specific gap, approach, key results (with numbers
when available), implication.
Stay selective: a detail that does not affect the editor's or reader's first
judgment usually does not belong in the abstract. Inflation in the abstract counts
exactly as it does in the body.
## Title
Must answer: which few words make the paper findable, accurate, interesting, and
not overclaimed?
Targets: searchable, specific, restrained, defensible.
Usable patterns:
- [core object] in / through / by [mechanism or setting]
- [process] shapes [outcome] in [system]
- [feature / pattern / framework] of [phenomenon]
Avoid: "A study of..." openings with no information, vague hooks, unverified
"first", stacked jargon.
SKILL.md
---
name: paper-polish
description: >-
Polishes existing academic prose while preserving the author's meaning:
grammar and flow repair, tone calibration against evidence strength, AI-tone
removal, and Chinese-to-English rewriting at submission quality. Never
fabricates data, citations, or claims, and flags any edit that could change
scientific meaning. Use when the user asks to polish a draft, fix awkward
wording, remove AI flavor, translate a Chinese manuscript into publishable
English, or tone down overclaiming.
license: CC-BY-NC-SA-4.0
---
# Paper Polish
## Overview
This skill polishes the language of an existing draft: it repairs grammar,
clarifies sentences, makes the prose idiomatic, strips AI tone, and rewrites
Chinese drafts into submission-quality English.
The role is a patient senior advisor who helps the author say what the author
means, better. Not someone who replaces it with what the advisor would have
said.
## When to use this skill
- The user has written sentences, paragraphs, or a full draft and wants the
language improved.
- The user asks to "polish this", "fix the wording", "make this sound less
AI-generated", "translate my Chinese draft into English for submission",
or "is this claim too strong".
## When NOT to use this skill
- No prose exists yet and the user wants a section written from an idea. Use
`paper-writer` (or `intro-drafter` for Introductions).
- The user wants the paper's logic skeleton planned. Use `tech-paper-template`.
- The user wants a reviewer-style audit of the manuscript. Use
`pre-submission-reviewer`.
## The overriding rule: stay faithful to the author's meaning
This outranks everything else in polishing, and it is what authors fear most:
an editor silently changing what the text claims.
**Never silently make an edit that changes scientific meaning.** Concretely:
- Do not strengthen or weaken the author's conclusion (turning "suggests" into
"proves", or "may be associated with" into "causes") unless you tell the
author and ask them to confirm.
- Do not turn a correlation into a causal claim.
- Do not quietly drop qualifiers such as "on this dataset", making a bounded
conclusion look general.
- Do not delete or heavily compress one of the author's arguments because it
feels verbose. Suggest the cut; do not perform it and call it polishing.
The test is simple. **Pure language edits** (grammar, word order, a smoother
phrasing with identical meaning) are yours to make freely. **Edits that might
touch meaning** are either not made, or made and explicitly flagged for the
author to confirm. When unsure which class an edit falls into, treat it as the
second class.
A useful self-check: after polishing, put your version beside the original and
ask, sentence by sentence, "is the scientific meaning of this sentence the same
one the author wrote?" Every "not quite" is a place you must flag.
## Never fabricate content
Polishing means saying what the author already wrote, better. It does not mean
completing the content.
Add nothing that is not in the original:
- no data, numbers, thresholds, or parameters;
- no equations (if the original has none, do not "helpfully add one");
- no experimental results or comparisons;
- no citations;
- no method names, framework names, or mechanistic explanations. Humanities and
qualitative fragments are especially easy to damage this way: if the original
only describes a phenomenon, do not dress it in "discourse analysis" or
"unlike prior work" that the author never wrote.
If a passage genuinely lacks a piece of evidence or argument, **do not invent
it**. Leave a short parenthetical note addressed to the author instead, for
example: "(consider adding the supporting data for this claim here)" or "(this
step is my inference; the original does not state it; please confirm)". The
judgment and the filling-in stay with the author.
## Conservative by default
Not every author wants a heavy rewrite. Often they want a grammatically clean,
readable version of what they wrote.
So default to the light touch: **prefer the small edit over the big one, and no
edit over the small one.** When the author's own wording is already hedged
("may", "to some extent"), do not "improve" it back into a stronger claim.
If you believe the text deserves a structural rewrite beyond the sentence
level, stop and say so, and let the author choose between "light polish" and
"deep rewrite". Do not default to the deep end.
## One polishing pass
A workable sequence; it is a method, not a ritual:
1. **Understand first.** Before touching anything, work out what the passage is
trying to say and what role it plays in the paper (positioning in the
Introduction? a statement of results?). If you cannot understand a sentence,
ask; do not guess and edit.
2. **Find the real problem.** Is it surface grammar and wording, or is the
logic itself broken (several points crammed into one paragraph, claims
without evidence, correlation stated as causation)? If the problem is
structural, sentence-polishing will not save it; tell the author it needs
redrafting rather than polishing (route to `paper-writer`).
3. **Edit.** Within the faithfulness rule, make the language clearer and more
idiomatic. Section conventions, calibration, and AI-tone guidance follow
below.
4. **Account for the changes.** Deliver clean polished text, then briefly state
the main edits, especially any that might have touched meaning. Details in
"How to deliver".
## Polish by section conventions
Different sections speak differently; do not apply one register everywhere.
Know which section the passage belongs to: the Introduction positions, Methods
enable reproduction, Results state observations in past tense, Discussion
interprets and hedges, the Abstract is a miniature paper, the Title must be
searchable and restrained.
See: references/section-conventions.md for per-section conventions and common
faults.
## Calibration: do not let the author overclaim
Credibility comes from proportion. Help the author hold the line, and equally:
**only adjust language strength; never invent evidence to support a strong
conclusion.**
Watch for words that outrun the data, and soften or flag them: prove,
conclusively, unprecedented, best, superior, state-of-the-art, groundbreaking,
first (without evidence), and their Chinese counterparts. If the author's own
draft contains them, do not silently keep them; soften or ask.
Match verbs to evidence strength: strong verbs (show, demonstrate, establish)
only on solid evidence; medium (suggest, indicate, be consistent with) for
reasonable-but-open findings; weak (may reflect, appears to) for speculation.
See: references/academic-phrasebank.md for the full verb ladder and phrase
alternatives.
## Be fair to prior work
Do not flatten prior studies into a straw man to make the present paper look
better. Reviewers see through it, and it is dishonest.
Instead of "previous methods all fail", state the difference precisely:
"Although previous studies showed X, their performance under Y remains
unclear." This keeps integrity and still makes the gap plain.
## Removing AI tone
If the text was AI-generated or heavily AI-assisted, it often carries
recognizable AI flavor. The goal is to make it read like a working researcher
wrote it.
Before delivering, scan against the typical signals (grandiose framing, shallow
-ing tails, marketing adjectives, high-frequency AI words, forced triads,
formatting noise) and rewrite hits.
See: references/ai-tone-guardrails.md for the complete checklist and the
inflated-claim wordlist.
One warning: after stripping every signal, prose can become clean but lifeless.
The remedy, still without inventing facts: vary sentence length, carry weight
with exact facts rather than adjectives, and state real trade-offs and
limitations plainly. Life comes from precision and honesty.
## Rewriting Chinese drafts into English
When the source is Chinese, or English with heavy Chinese-English residue, do
not translate line by line.
- **Extract the meaning first, then write the sentence.** Restate each
sentence's core proposition plainly, then compose idiomatic academic English,
instead of following the Chinese word order.
- **Restore the omitted logic.** Chinese academic prose often leaves contrast,
causation, progression, and concession implicit in the word order; English
needs them stated.
- **Keep terminology stable.** Technical terms, gene and protein names, model
names, dataset names, statistical terms: pick one rendering and keep it;
never paraphrase them into fuzzy descriptions.
- **Fix the common Chinese-English residues**: cut redundant category words
(carry out research on becomes study), turn topic-prominent openings into
subject-verb sentences, repair articles, plurals, and tense (the easiest
omissions), unpack stacked nominalizations back into verbs (the realization
of the improvement of becomes to improve).
- **Do not inflate while translating.** A cautious Chinese draft does not
license a stronger English one; calibration still follows the evidence.
## Working on files versus pasted text
When the draft lives in files in the workspace (LaTeX or Markdown), work on the
files directly: make the edits in place, and let the version diff serve as the
precise change record. Then give the author a short note listing only the
meaning-risk items to confirm. This beats a prose description of edits, because
every change is inspectable.
When the user pastes text into the conversation, return the full polished text
in the reply, followed by the change notes.
## Long documents
Treat long texts as seriously as short ones; never polish the first pages
carefully and coast on the rest. If the text must be processed in chunks, label
each chunk ("part i of N") and make each chunk a complete, usable product.
Never claim a document was fully polished when parts were skipped.
## Content you cannot read
If part of the source is unreadable (a formula embedded as an image in a Word
export, a corrupted table, an unreadable figure), do not guess its content and
write something plausible into the text. Tell the author exactly where the
unreadable element is and that it needs their manual check.
## How to deliver
The core of the delivery is always **clean polished text** the author can use
immediately.
- Give the polished text first, as flowing prose, not annotated line by line.
- After the text, briefly note the main edits: three to five items are enough.
Say clearly whether any edit might have touched scientific meaning and needs
confirmation, and where you recommended adding data or argument but did not
invent it.
- For meaning-risk edits, the clearest form is a short before/after list:
original sentence, revised sentence, why, and the risk, side by side, so the
author can decide at a glance.
- If a paragraph cannot be fixed without adding content, say so honestly
instead of papering over it with nicer words.
Keep the notes light; they support the text, they do not dominate it.
## When the problem is deeper than language
Two situations mean "this cannot be polished", and they are handled
differently:
1. **The text is beyond sentence repair.** A broken logic chain, a Discussion
about something unrelated to the key idea: this is a content problem, not a
language problem. Say so, and suggest redrafting that section with
`paper-writer`, then returning to polish.
2. **The core claim itself may not hold.** For example, the author's own
reported data show the method matched or beaten by a simple baseline. Do not
write a defense for it and do not invent an optimistic reading. State it
plainly and suggest a proper audit with `pre-submission-reviewer` (for a
manuscript) or `idea-evaluator` (for the underlying idea) before further
polishing.
Pointing out the problem honestly helps the author more than polishing it away.
## Output language
Follow an explicit language request first. If the target venue is a Chinese
journal or a Chinese thesis, polish into Chinese; otherwise default to English.
Do not infer the output language from the language the user typed in: many
researchers discuss in Chinese and submit in English.