PaperCash plugin
Searches Semantic Scholar, arXiv and Crossref, drafts an abstract-level review and formats DOI citations in GB/T 7714 and APA
- 159 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- The script sends search queries to external paper databases and uses your access rights when a proxy or cookie is configured
- The call log on disk records commands and text length, and exports may contain personal data
Install
- Works in
Manual install
Clone https://github.com/Jesseovo/PaperCash and read SKILL.md. Create a separate Python 3.10+ environment, install `requirements.txt`, run `python scripts/papercash.py --diagnose` and show the result. Then ask the user which topic to research. Add source keys and cookies only if the user provided them.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The script sends search queries to external paper databases and uses your access rights when a proxy or cookie is configured
- The call log on disk records commands and text length, and exports may contain personal data
This is third-party code. Review the repository files before installing.
What it does
PaperCash is a Python tool for research work with an agent. The `scripts/papercash.py` script searches Semantic Scholar, arXiv, Crossref and Baidu Scholar by topic and returns title, authors, year, DOI, link and abstract. PubMed is enabled with `--sources pubmed`, and Google Scholar, CNKI and Wanfang join the default search once a proxy or cookie is configured. From the metadata it finds, the `review` command drafts a literature review and can export it to Word. The `cite` command turns a DOI into a citation in GB/T 7714, APA, MLA, Chicago or BibTeX. Smaller helpers cover an outline draft, ideas for expanding a paragraph, academic wording edits and a check of a Word file against common format samples. The `check` and `humanize` commands give limited results. The first compares sentences with abstracts of found papers and answers "cannot judge" when there are no candidates; the second proposes rewordings and asks for confirmation. Both write the time, command and input length, not the text itself, to `~/.papercash/audit/`.
Who it is for. Students and postgraduates writing term papers and theses in Chinese or English, and researchers assembling a reference list for an article.
Good fit when
- You need a dozen recent papers on a topic with DOIs and source links quickly
- A reference list has to be formatted in GB/T 7714 or APA from a set of DOIs
- Before reading you want an abstract-level overview to decide which papers to open in full
- A Word file goes to the department, and it helps to compare its layout with a common sample first
Not a fit when
- You need a full literature review: `review` relies on metadata and abstracts only, so methods and conclusions still need the original papers
- You need an official plagiarism report: `check` is a preliminary comparison and the university runs the real one
- You need to predict how a text will do against an AI detector: `humanize` does not promise that
- Your department formats differently from the common Word check samples
Example request
Find ten recent papers on medical image segmentation, keep the DOIs and links, then format them in GB/T 7714Limitations
Python 3.10 or newer is required, with dependencies from `requirements.txt` installed in a separate environment. Source availability depends on the network and API limits. Google Scholar, CNKI and Wanfang work only after a proxy or cookie is set up, and you should use your own access rights.
A zero-result search and a source failure are different outcomes: the first means no papers, the second means the source did not answer. Database metadata is often incomplete or duplicated, so check DOIs, authors and years by hand before submission. Citation support covers common styles, and the requirements of a university or journal take priority.
The `check` and `humanize` scripts keep a call log in the home folder, and `check` compares text only with the abstracts it found.
How to disable. Delete the `papercash` skill folder from the agent skills directory (`.claude/skills`, `.agents/skills` or wherever you copied it). The call log is in `~/.papercash/audit/` and settings in `~/.config/papercash/.env`; delete them by hand.
FAQ
Which sources does PaperCash use?
Semantic Scholar, arXiv, Crossref and Baidu Scholar are on by default. PubMed is enabled with `--sources pubmed`. Google Scholar, CNKI and Wanfang join after a proxy or cookie is configured.
How do I install PaperCash in Claude Code?
The repository ships `.claude-plugin/plugin.json` and `SKILL.md`. Clone it, create a Python environment, install `requirements.txt` and ask the agent to follow SKILL.md.
Can I trust the `check` result?
No, it is a preliminary comparison. The command matches sentences with abstracts of found papers, answers that it cannot judge without candidates, and a similarity figure proves neither plagiarism nor originality.
Which citation formats does PaperCash support?
GB/T 7714, APA, MLA, Chicago and BibTeX. The `cite` command builds a citation from a DOI, and `review` with `--export-docx` writes the review to Word.
What does PaperCash store on disk?
A log in `~/.papercash/audit/` with the time, command and input length without the text itself, plus a search cache and your settings in `~/.config/papercash/.env` if you created them.
Discussion
No questions or reviews yet.
Related
PluginsA skill that searches five open databases for papers and builds an offline report with relevance, BibTeX and search strategies.
Checks every reference of a paper against Crossref, dblp and arXiv before submission, with no language model in the verdicts.
The paper-spine skill takes a topic, materials and data through a scientific paper: literature, text, figures, review and Word, LaTeX and PDF files
A skill set for academic work: paper review, pre-analysis plan check, grant proposal review, code audit and slide decks.
Takes an agent from a topic to an arXiv review paper in LaTeX IEEEtran: approved plan, every citation checked against a live source, PDF build
Eight skills for LaTeX research papers, from topic to compiled PDF, with verified BibTeX citations and gated steps
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Repository stars | Main difference |
|---|---|---|---|
| PaperCash (this card) | no | 159 | |
| paper-preflight | no | 182 | checks a manuscript before submission and has no paper search |
| Nature-Paper-Skills | no | 671 | helps write papers for Nature-group journals |
README in short
The repository ships a skill and CLI for working with research papers: search across several databases, an abstract-level review, citation formatting and auxiliary checks. The author states the limits outright: search results do not replace reading the full text, and similarity is not a university check result. A page with a daily refreshed list of recent papers sits next to it.
README badge
Original SKILL.md
--- name: papercash description: 使用本地 Python CLI 辅助论文检索、摘要级文献整理、DOI 引用格式化、Word 格式检查、查重预检和写作修改。涉及“降AI率”时只提供表达建议；不把检索结果当成全文阅读或正式查重结果。 metadata: version: "2.0.0" --- # PaperCash 本 Skill 的入口是 `scripts/papercash.py`。先在 Skill 目录安装 `requirements.txt`，然后按用户实际请求选择命令。无需为了普通论文问题强制运行全部功能。 ## 命令 | 任务 | 命令 | | --- | --- | | 检索论文 | `python scripts/papercash.py search "主题" --emit agent --limit 20` | | 整理文献 | `python scripts/papercash.py review "主题" --format gb7714` | | 起草大纲 | `python scripts/papercash.py outline "题目"` | | 获取扩写思路 | `python scripts/papercash.py expand "段落"` | | 修改表达 | `python scripts/papercash.py polish "段落"` | | 相似内容预检 | `python scripts/papercash.py check "文本或路径" --emit agent` | | 表达修改建议 | `python scripts/papercash.py humanize "文本或路径" --emit agent` | | DOI 引用 | `python scripts/papercash.py cite "10.xxxx/..." --style gb7714` | | Word 格式检查 | `python scripts/papercash.py format "论文.docx"` | | 数据源诊断 | `python scripts/papercash.py --diagnose` | `search` 支持 `--sources`、`--export-docx`、`--save-dir`。`review` 支持 `--format gb7714|apa`、`--export-docx`、`--save-dir`。查重预检支持 `--mode fast|standard|deep|legacy`。确切参数以 `--help` 为准。 ## 使用结果时的判断 - 逐条核对论文标题、作者、年份、DOI 与链接。数据库元数据可能不完整或重复，引用格式输出也需要人工核对。 - `review` 依据检索到的元数据与摘要生成整理稿。涉及方法、结论、研究空白时，阅读原文后再作判断；不要将其表述为已经完成全文综述。 - `check` 仅将句子与检索到的标题和摘要比较。没有候选文献时报告“无法判断”，任何相似度都不能证明抄袭或原创。正式查重使用学校指定系统。 - `humanize` 只提供有限的表达修改建议，不预测 AI 检测结果。工具交互式要求确认；默认保留识别到的数据、公式、法条和引用所在句。修改后仍需核对原意。 - `check` 与 `humanize` 会在本机 `~/.papercash/audit/` 记录时间、命令、输入摘要和长度，不保存原文。 - `expand` 输出扩写思路和模板，不会凭空补充可靠数据或文献。`format` 使用固定的常见格式样例，具体要求以学校模板为准。 ## 交付方式 把命令结果整理成用户能直接使用的形式，保留可核对的来源链接，明确哪些内容仍需全文核实。如果数据源失败或没有结果，如实说明，不补造论文、引用、查重率或检测准确率。