Agent skill
search
Search academic papers in the local ScholarAIO knowledge base. Supports unified search (keyword + semantic fusion), keyword-only (FTS5), semantic-only (FAISS), author search, and federated search across main library, explore databases, and arXiv. Use when the user wants to find papers, look up literature, search by author, explore research topics, or search across multiple sources. For citation rankings and citation count updates, see the /citations skill.
Install this agent skill to your Project
npx add-skill https://github.com/ZimoLiao/scholaraio/tree/main/.claude/skills/search
SKILL.md
文献搜索
在本地论文库中搜索文献。默认使用融合检索(关键词 + 语义向量合并排序),也支持单独使用某一种模式。
执行逻辑
-
解析用户输入,判断搜索模式:
- 如果用户明确要求"语义搜索"、"向量搜索"或"vsearch",使用
vsearch - 如果用户明确要求"关键词搜索"、"全文搜索"或"FTS",使用
search - 如果用户明确按作者搜索(如"找某某的论文"、"某某发表的"),使用
search-author - 如果用户要求按引用量排序(如"引用最高的"、"最经典的"、"top cited"),转交
/citationsskill - 默认使用
usearch(融合检索)——同时执行 FTS5 关键词搜索和 FAISS 语义搜索,合并去重排序。两路都命中的论文排名靠前。向量索引不可用时自动降级为纯关键词。 - 如果用户要求跨库搜索(如"也搜一下 arXiv"、"在 explore 库里也找找"、"也搜 proceedings"、"全部来源"、"联邦搜索"),使用
fsearch
- 如果用户明确要求"语义搜索"、"向量搜索"或"vsearch",使用
-
从用户输入中提取:
- 查询词:用户想搜索的内容
- 返回数量:用户指定的
--top N,未指定则使用默认值 - 年份过滤:
--year 2023(单年)、--year 2020-2024(范围)、--year 2020-(起始年至今) - 期刊过滤:
--journal "Fluid Mechanics"(模糊匹配) - 类型过滤:
--type review(模糊匹配,常见值:review、journal-article、book-chapter)
-
执行搜索命令:
融合检索(默认):
scholaraio usearch "<查询词>" --top <N> [--year <Y>] [--journal <J>] [--type <T>]
关键词搜索:
scholaraio search "<查询词>" --top <N> [--year <Y>] [--journal <J>] [--type <T>]
语义搜索:
scholaraio vsearch "<查询词>" --top <N> [--year <Y>] [--journal <J>] [--type <T>]
作者搜索:
scholaraio search-author "<作者名>" --top <N> [--year <Y>] [--journal <J>] [--type <T>]
引用量排序:使用
/citationsskill 中的scholaraio top-cited命令。
联邦搜索(跨库 + arXiv):
# 同时搜主库和 arXiv
scholaraio fsearch "<查询词>" --scope main,arxiv --top <N>
# 同时搜主库和 proceedings
scholaraio fsearch "<查询词>" --scope main,proceedings
# 同时搜主库和所有 explore 库
scholaraio fsearch "<查询词>" --scope main,explore:*
# 搜指定 explore 库
scholaraio fsearch "<查询词>" --scope explore:my-survey
# 仅搜 arXiv(在线查询,不需要本地数据)
scholaraio fsearch "<查询词>" --scope arxiv
# 全部来源
scholaraio fsearch "<查询词>" --scope main,proceedings,explore:*,arxiv
--scope 支持逗号分隔组合:main(主库融合搜索)、proceedings(论文集子论文)、explore:<名称> 或 explore:*(explore 库)、arxiv(在线 arXiv API)。默认 scope 为 main。arXiv 结果会标注 [已入库] 表示该论文已在本地库中。
-
将搜索结果整理后呈现给用户。融合检索结果中每项标注了匹配来源:
both:关键词和语义都命中(最相关)fts:仅关键词命中vec:仅语义命中
-
复杂查询:当 CLI 参数组合无法满足需求时(如按一作姓氏首字母筛选、多条件交叉、自定义排序等),直接写 Python 读
data/papers/*/meta.json做查询。JSON 关键字段:
title, authors, first_author, first_author_lastname, year, doi, journal,
abstract, paper_type, citation_count (dict: crossref/semantic_scholar/openalex),
ids, toc, l3_conclusion
示例
用户说:"帮我搜一下 turbulent boundary layer 相关的论文"
→ 执行 usearch "turbulent boundary layer"
用户说:"用语义搜索找 drag reduction 的文献,给我前5篇"
→ 执行 vsearch "drag reduction" --top 5
用户说:"找 Liao Z-M 的论文"
→ 执行 search-author "Liao"
用户说:"我库里引用最高的论文有哪些"
→ 转交 /citations skill(使用 top-cited 命令)
用户说:"2020年以后关于 drag reduction 的论文"
→ 执行 usearch "drag reduction" --year 2020-
用户说:"JFM 上发的湍流论文"
→ 执行 usearch "turbulence" --journal "Fluid Mechanics"
用户说:"库里引用最高的 review 文章"
→ 转交 /citations skill(使用 top-cited --type review 命令)
用户说:"帮我在 arXiv 上也搜一下 physics-informed neural network"
→ 执行 fsearch "physics-informed neural network" --scope main,arxiv
用户说:"所有来源都搜一下 drag reduction,包括 explore 库"
→ 执行 fsearch "drag reduction" --scope main,proceedings,explore:*,arxiv
用户说:"在我之前建的 wall-bounded-turbulence explore 库里搜 channel flow"
→ 执行 fsearch "channel flow" --scope explore:wall-bounded-turbulence
用户说:"连 proceedings 一起搜 granular damping"
→ 执行 fsearch "granular damping" --scope main,proceedings
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?