Agent skill
citation-check
Verify citations in AI-generated or human-written text against the local knowledge base. Catches hallucinated references, wrong metadata, and missing papers. Use when the user wants to check if citations are real and accurate.
Install this agent skill to your Project
npx add-skill https://github.com/ZimoLiao/scholaraio/tree/main/.claude/skills/citation-check
SKILL.md
引用验证
检查文本中的引用是否真实、准确,防止 AI 幻觉引用和元数据错误。
重要背景:AI 生成的学术文本中约 40% 的引用可能是幻觉(编造的论文、张冠李戴、元数据拼凑)。即使是人类写的文本也常有年份/期刊名错误。本 skill 的目标是在投稿前消灭所有引用问题。
CLI 命令(首选方法)
# 检查文件中的引用
scholaraio citation-check <file>
# 在工作区范围内验证
scholaraio citation-check <file> --ws <workspace-name>
# 从 stdin 读取
cat draft.md | scholaraio citation-check
CLI 自动提取 author-year 格式的引用(Author (Year)、(Author, Year)、Author & Author (Year)、Author et al. (Year) 等),在本地知识库中搜索匹配,输出每条引用的验证状态:
| 状态 | 含义 |
|---|---|
| VERIFIED | 本地库有唯一匹配,作者+年份一致 |
| AMBIGUOUS | 本地库有多条匹配,需人工确认 |
| NOT_IN_LIBRARY | 本地库中无此论文 |
对于 AMBIGUOUS 和 NOT_IN_LIBRARY 的结果,可进一步执行手动深度验证(见下文)。
前提
用户提供待检查的文本(粘贴、文件路径、或指定 workspace 中的草稿文件)。 如有 workspace,优先在工作区范围内验证。
手动深度验证
CLI 完成初步筛查后,对问题引用执行更细致的检查:
Layer 1 — 本地库匹配
scholaraio search-author "<Author>" --top 5
scholaraio usearch "<关键词 from title>" --top 5
在本地库中找到匹配论文后,核对:作者名、年份、标题、期刊是否一致。
Layer 2 — DOI/元数据核验
如果本地库有匹配,读取 meta.json 中的 DOI 和详细元数据交叉比对。 如果本地库无匹配,提醒用户——该引用不在工作区/知识库中。
Layer 3 — 内容一致性
对于关键引用(支撑核心论点的),加载 L2-L3 检查:
scholaraio show <paper-id> --layer 3
验证:文本中对该论文的描述是否与论文实际内容一致?是否存在过度解读或断章取义?
常见问题模式
- AI 幻觉引用:作者名和年份拼凑出一篇不存在的论文——标记 SUSPICIOUS
- 张冠李戴:引用了真实论文但描述的是另一篇的内容——标记 CONTENT MISMATCH
- 元数据错误:年份差一年、期刊名拼错、一作搞错——标记 METADATA MISMATCH 并给出正确值
- 过度引用:一个论点堆了 5+ 引用但大部分并不直接相关——建议精简
示例
用户说:"帮我检查这段文字里的引用是否正确"
→ 先运行 scholaraio citation-check <file>,再对问题引用逐条深入验证
用户说:"检查 workspace/my-paper/introduction.md 里的引用"
→ 运行 scholaraio citation-check workspace/my-paper/introduction.md --ws my-paper
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
lammps
Use when working on classical materials simulations with LAMMPS, especially interatomic-potential selection, shock or deformation setups, thermodynamic runs, and structure analysis for solids or nanomaterials.
citations
View top-cited papers ranking and refetch citation counts from APIs. Use when the user asks about highly cited papers, citation rankings, or wants to update citation data.
paper-writing
Assist with writing sections of a research paper (Introduction, Related Work, Method, Results, Discussion, Conclusion). Leverages workspace papers for citations and evidence. Use when the user wants help drafting or revising specific paper sections.
arxiv
Use when the user wants to browse arXiv preprints, search arXiv directly, fetch a PDF by arXiv ID or URL, or send a preprint straight into the ingest pipeline.
bioinformatics
Use when working on bioinformatics toolchains such as alignment, variant calling, phylogenetics, or protein-structure analysis, especially when the agent must route across BLAST, minimap2, samtools, bcftools, MAFFT, IQ-TREE, or ESMFold.
review-response
Draft point-by-point responses to peer review comments. Locates supporting evidence from workspace papers and the original manuscript. Use when the user receives reviewer feedback and needs to write a rebuttal or revision response letter.
Didn't find tool you were looking for?