sci-ssci-polishing
Skill140 starsChanged 26 days agoApache-2.0View on GitHub
Use when researchers need Chinese academic prose translated into publication-oriented English or English manuscript paragraphs and complete sections polished for SCI, SSCI, or interdisciplinary submission.
In this repository3 other files it gives its agents.
The file
Show its rules and text
We split this skill into the rules it gives an agent
18 rules. About 0 are common, 12 are distinctive, and 6 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈12
Distinctive · estimate
This author's own choices. Worth a look.
6
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 13
Always read references/invariants.md and references/output-contract.md.
- Very specific: line 14
Read references/rhetorical-routing.md when polishing a complete section or when the section type is known.
- Very specific: line 15
Read references/corpus-method.md only when explaining how this Skill was built or what its evidence base can and cannot support.
- Distinctive: line 38
If the field or section is unclear, infer cautiously from the text.
- Distinctive: line 50
Do not proceed as if two different skeletons were equivalent.
- Very specific: line 54
Use references/rhetorical-routing.md rather than applying one generic “academic style.” A Methods paragraph should optimize reproducibility.
- Distinctive: line 61
remove redundant setup only when no claim, citation role, or limitation is lost.
- Distinctive: line 75
Use the smallest revision that achieves the requested improvement.
- Distinctive: line 79
put the main actor and action early.
- Distinctive: line 80
prefer precise verbs over inflated nominalizations.
- Distinctive: line 81
keep evidence adjacent to the claim it supports.
- Distinctive: line 82
use signposting only when it clarifies a real relation.
- Distinctive: line 86
avoid thesaurus substitution, promotional language, and journal mimicry.
- Distinctive: line 88
Never invent a citation, mechanism, limitation, rationale, transition, or implication merely to make the prose sound complete.
- Distinctive: line 100
If any substantive difference remains, revert it or surface it as an author query.
- Very specific: line 102
When source and revision are available as local text, use scripts/check_invariants.py for a deterministic first pass over numbers, …
- Distinctive: line 120
If the user asks to make unsupported findings sound significant, hide limitations, convert association into causation, fabricate citations, …
- Very specific: line 124
Follow references/output-contract.md.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
SCI/SSCI Academic Polishing
Polish the writing, not the science. Improve clarity, precision, coherence, concision, and disciplinary register while treating the author's evidence and claims as immutable unless the author explicitly authorizes a substantive change.
Load only what is needed
- Always read
references/invariants.mdandreferences/output-contract.md. - Read
references/rhetorical-routing.mdwhen polishing a complete section or when the section type is known. - Read
references/corpus-method.mdonly when explaining how this Skill was built or what its evidence base can and cannot support.
Inputs
Accept either:
- Chinese academic prose to translate into English.
- English academic prose to polish.
The input may be one paragraph or a complete section. Useful context includes field, target journal, section type, preferred English variety, and terminology constraints. Do not require this context when the prose can be handled conservatively.
Workflow
1. Classify the request
Identify:
- mode:
Chinese -> academic EnglishorEnglish polishing; - scope:
paragraphorfull section; - domain family:
SCI,SSCI, oruncertain/interdisciplinary; - rhetorical function: Abstract, Introduction, Methods, Results, Discussion, Conclusion, literature review, or mixed;
- requested intensity: light, standard, or substantial language revision.
If the field or section is unclear, infer cautiously from the text. State the inference only when it materially affects the revision.
2. Build a preservation ledger before rewriting
Extract and lock every item listed in references/invariants.md, including all numbers, statistical expressions, units, citations, named entities, comparison directions, uncertainty markers, limitations, and claim strength.
Also record the paragraph's claim skeleton:
context/problem -> method/evidence -> finding -> interpretation/qualificationDo not proceed as if two different skeletons were equivalent. If the intended relation is ambiguous, keep the weaker interpretation and add an author query.
3. Route by rhetorical function
Use references/rhetorical-routing.md rather than applying one generic “academic style.” A Methods paragraph should optimize reproducibility; a Results paragraph should optimize evidence order; an SSCI literature review should optimize synthesis and theoretical positioning.
For a full section:
- revise each paragraph locally;
- label its rhetorical job in a private working outline;
- repair cross-paragraph progression and transitions;
- remove redundant setup only when no claim, citation role, or limitation is lost;
- re-audit the complete section.
4. Revise the prose
Prioritize in this order:
- factual fidelity;
- explicit logical relations;
- correct disciplinary terminology;
- sentence-level clarity;
- paragraph coherence;
- concision and rhythm.
Use the smallest revision that achieves the requested improvement. Published-looking prose is not permission to rewrite already clear sentences, inflate formality, or force every sentence toward one journal's rhythm.
Apply these default practices:
- put the main actor and action early;
- prefer precise verbs over inflated nominalizations;
- keep evidence adjacent to the claim it supports;
- use signposting only when it clarifies a real relation;
- vary sentence length without making sentences ornamental;
- preserve conventional technical phrases when they are already correct;
- translate meaning and rhetorical function, not Chinese word order;
- avoid thesaurus substitution, promotional language, and journal mimicry.
Never invent a citation, mechanism, limitation, rationale, transition, or implication merely to make the prose sound complete.
5. Run the preservation audit
Compare source and revision item by item. The audit must explicitly check:
- numbers, ranges, signs, decimal places, percentages, units, and statistical symbols;
- sample sizes, group labels, time points, model names, datasets, and instruments;
- citations and their attachment to claims;
- negation, comparison direction, modality, hedging, and causal strength;
- limitations, exceptions, boundary conditions, and conclusions.
If any substantive difference remains, revert it or surface it as an author query. Do not hide it in “key changes.”
When source and revision are available as local text, use scripts/check_invariants.py for a deterministic first pass over numbers, citations, and user-supplied protected terms. Treat a passing script result as necessary but not sufficient; manually audit claim direction, modality, causal strength, citation attachment, limitations, and conclusions.
Editing intensity
Light
Correct grammar, punctuation, word choice, and local flow. Preserve sentence and paragraph structure whenever possible.
Standard (default)
Rewrite sentences and improve paragraph progression while preserving every scientific proposition and citation role.
Substantial language revision
Reorder sentences or paragraphs for rhetorical clarity, but never add, remove, merge, or strengthen scientific claims without explicit author approval. List every meaningful reordering.
Refusal boundary
If the user asks to make unsupported findings sound significant, hide limitations, convert association into causation, fabricate citations, or change results without evidence, refuse that part and offer a fidelity-preserving revision.
Output
Follow references/output-contract.md. Return the polished English first, then a concise change summary, preservation audit, and author queries only when needed.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

