paper-skill: evidence-first manuscript workflow
paper-skill
A skill for researchers who draft in Chinese: drafts and results become an English manuscript, with citations and claims checked
- 143 ★ on the repository
- Last commit
Minimal access
Why this level
- The skill reads an unpublished manuscript and data: upload only what is needed and strip sensitive details first
- Installation uses a `curl | bash` script that you should read before running
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add skills/paper-skillDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewMinimal accessThe tool runs nothing and makes no network calls.
- The skill reads an unpublished manuscript and data: upload only what is needed and strip sensitive details first
- Installation uses a `curl | bash` script that you should read before running
This is third-party code. Review the repository files before installing.
What it does
paper-skill carries a manuscript from a Chinese draft, result tables, figures and a reference list to an English version for a journal. It builds a claim-to-evidence map, writes an abstract, main text and cover letter, then checks citations, figures and reviewer responses for consistency. Claims receive the status `VERIFIED`, `UNVERIFIED` or `AUTHOR_INPUT_NEEDED`, and the whole manuscript gets `READY`, `READY_WITH_WARNINGS` or `NOT_READY`. The skill explicitly refuses to invent data, experiments, DOI values, journal rules or revision locations. There are three main routes: `write` for the text, `review` for a pre-submission check and `submit` for rebuttals, the cover letter and the final package. A separate `paper-system` mode walks through eight steps with 88 prompts, from the storyline to shortening the text. For IEEE T-RO the repository has a dated venue profile.
Who it is for. Researchers in technical universities and labs who think and write in Chinese and submit to an English-language journal such as IEEE T-RO.
Good fit when
- A Chinese draft and result tables must become an English abstract and introduction
- A pre-submission check is needed: which claims lack data support and which citations look doubtful
- Reviewers sent comments and you need a response matrix with exact revision locations
- The cover letter and journal declarations should be built from verified material
- You want to follow the eight steps from storyline to shortening the text and resume at any step
Not a fit when
- You want a prediction of acceptance: the project explicitly says it does not predict editorial decisions
- The manuscript is in Russian: the skill is tuned for Chinese source materials
- You work with confidential data and cannot strip it out first: sensitive details must be removed before upload
- The text is for daily work or contracts: such prompts are excluded from the set
Example request
Use paper-skill. My manuscript is in Chinese and targets IEEE T-RO. Build a claim-to-evidence map first, then rewrite the abstract and show the submission blockers.Limitations
The skill is built for Chinese source material: the 88 prompts come from the author's Chinese HTML, and a journal profile exists only for IEEE T-RO, so for other venues you have to supply the requirements. Historical examples do not count as evidence of current journal rules.
The installer looks for agent platform directories on the machine and defaults to Codex if none exist, and it needs Git. The Python verification scripts run separately and require version 3.10 or newer. The public test of 8 cases checks only safety behaviour and does not measure text quality.
How to disable. Delete the `paper-skill` folder from your agent's skills directory, for example `~/.codex/skills`. The installer creates no services or keys.
FAQ
How do I install paper-skill?
On macOS and Linux run `curl -fsSL https://raw.githubusercontent.com/cLin-c/paper-skill/main/install.sh | bash`, and on Windows run `install.ps1` from the same repository. The installer uses agent directories it finds, or sets up the Codex variant.
What do VERIFIED and UNVERIFIED mean?
`VERIFIED` goes on a claim backed by data or a source in the author's materials. `UNVERIFIED` means there is no backing, and `AUTHOR_INPUT_NEEDED` asks the author for the missing fact.
Does the skill write the paper for me?
It prepares English text from your materials and assembles a submission package, but adds no data absent from your sources. Decisions about conclusions and the final check against journal rules stay with the author.
Which journals are supported?
The repository has a ready profile only for IEEE T-RO, dated against official sources. For another journal, supply its requirements, and the `journal` route helps check the strategy.
Does the skill work in Claude Code?
The project description calls it a Claude Code skill, and the repository also has an `agents/openai.yaml` file for Codex. The shared entry point is `SKILL.md`, which both agents pick up.
Discussion
No questions or reviews yet.
Related
SkillsA Claude Code skill for research papers: a five-stage pipeline, section rules, a style audit and an independent red-team review.
42 skills for the whole research paper path: from discovery and drafting to citation checks, submission, rebuttal and publication
A skill set for academic papers: deep method analysis, visual slides and a coherent write-up from a paper PDF
Skill with 30 prompt templates for research papers in English and Chinese, from literature review to grant proposals
A skill that drafts an ML paper after the patterns of CVPR, ICCV and NeurIPS best papers and draws its figures and schemes.
Ten Codex skills for IEEE papers: writing, reviewer-style checks, experiments, figures, LaTeX and responses to reviews.
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| paper-skill: evidence-first manuscript workflow (this card) | no | 143 | |
| Research Paper Writing | no | 8,297 | a general paper-writing skill without evidence statuses |
| Academic Research Skills | no | 56,682 | a set of research skills without a manuscript submission pipeline |
README in short
The README details the routes and the `paper-system` mode with its eight steps. It includes a public set of 8 test cases, examples and safe prompts for a first run. The `tools/paper_skill_gate.py` and `tools/run_full_workflow.py` scripts check a submission package against config files.