nature-academic-search Skill for Claude: Literature Search, Citation Checks and Export

nature-academic-search makes Claude search papers across sources, check citations and export references, using a local search server or built-in scripts.

Installs: 15.8K; on skills.sh
Stars: 51.7K; on GitHub
Files: 53; besides SKILL.md
Updated: Nov 2026; last repo commit

nature-academic-search is a router skill in the nature-skills collection by Yuan1z0825. Claude works out which of six workflows you need (multi-source search, citation verification, MeSH strategy, citation-file conversion, reference management or a strict external-citation audit) and loads only that workflow's guide.

Search uses a bundled MCP server that queries CrossRef, PubMed and arXiv, with Scopus and ScienceDirect as opt-in sources. Without the server, two bundled Python scripts search OpenAlex and convert DOIs, PMIDs or arXiv IDs to .ris, .bib, .enw and .nbib files.

The repo README marks this skill Beta, meaning tested on examples with edge cases possible. PubMed search needs an email address in configuration, and Scopus and ScienceDirect need your own Elsevier access.

Good for

Researchers who need a deduplicated paper table with DOIs, authors and sources
Biomedical authors building MeSH-assisted PubMed searches
Authors who convert reference lists to RIS, BibTeX or other citation styles
Authors auditing strict external citations and influential citers of their own paper

Install nature-academic-search

Claude Code

Copy this prompt.

Install the nature-academic-search skill in Claude Code. Run this command:

npx skills add https://github.com/yuan1z0825/nature-skills --skill nature-academic-search

Then check that it loaded and tell me how to use it.

2
Paste it into Claude Code. It shows the change and asks before running it.

Requires Node.js.

From its listing on skills.sh. Source.

What's in the skill

When it loads

Search literature across sources, verify or manage citations, and build MeSH strategies or citation-impact audits. Use for 文献检索、引文核对、参考文献管理、严格他引 and evidence-backed citer profiles; not for translating a paper or drafting manuscript prose.

Sections in SKILL.md

Routing protocol
1. Load the manifest and the core layer
2. Detect the workflow
3. Load the matching workflow fragment(s)
4. Run the workflow using the loaded material
5. Reach for references only when needed

Key facts

License (per GitHub): Apache-2.0
Author: yuan1z0825 (repo owner)
Repository: yuan1z0825/nature-skills, folder skills/nature-academic-search
Installs on skills.sh: 15,778 as of 2026-11-02
Repo stars: 51,696 as of 2026-11-02 (whole repo)
Last repo commit: 2026-11-01
Files in the skill folder: 54 files including SKILL.md
Status in the repo README: Beta (tested on examples; edge cases possible)
skills.sh audits: Gen Agent Trust Hub: pass; Socket: pass; Snyk: pass

What it includes

Find literature across sources
Searches CrossRef, PubMed and arXiv by default and returns a deduplicated table with title, authors, year, journal, DOI and source.
Network · multi-source-search
Verify citations extracted from a document
Verifies citations pulled from a document, and the output lists records that could not be verified.
Network · citation-verification
Build a MeSH and PubMed search strategy
Adds MeSH terms and synonyms to a PubMed query.
Network · mesh-strategy
Convert .nbib, .ris and .bib files
Converts complete supplied records locally. SKILL.md says this does not need source lookup, API setup or a network check.
Files · citation-file-mgmt
BibTeX, related articles, ID conversion
Reference management and related-article discovery.
Network · reference-mgmt
Find independent citations and influential citers
Determines strict independent other-citations, builds article-level citation metric tables, identifies high-profile citers and extracts how they cited the paper.
Network · strict-other-citation-impact-audit
Search, details, citation formatting, MeSH, Scopus, ScienceDirect
Tools include search_papers, get_paper_by_id, get_citation, lookup_mesh and a set of Scopus and ScienceDirect tools. Scopus and ScienceDirect run only when you ask for them.
Network · MCP search server
No-MCP discovery search via OpenAlex
Queries the OpenAlex API directly. An OpenAlex API key is optional and raises the daily budget.
Network · scripts/academic_search.py
Turn DOIs, PMIDs or arXiv IDs into reference files
Fetches records from CrossRef, PubMed and arXiv, then writes .ris, .bib, .enw or .nbib files.
Files · scripts/format-converter.py
Check that the API endpoints are reachable
Requests PubMed, CrossRef and arXiv test URLs before batch work.
Network · scripts/preflight.py

From the official docs

Permissions

Medium risk

This skill sends your search terms to scholarly APIs, runs a local Python search server and writes reference files, and a bundled installer can edit your Claude settings. Install the server yourself and keep keys in local config.

What it can do

Read your data
Create and change things
Delete things
Read-only mode

Watch out for

The folder bundles an installer that edits Claude settings
install.sh runs pip install, copies the server into ~/.claude/mcp_servers and adds it to ~/.claude/.mcp.json and settings.json. Read it before you run it.
The search server downloads and runs Python packages
The sample config starts it with uv and pulls mcp, requests, toml, lxml and pybliometrics. Install from requirements.txt in an environment you control.
Optional keys and Firecrawl go to outside services
Scopus and ScienceDirect use your Elsevier quota. OPENALEX_API_KEY is sent to OpenAlex as a Bearer header. If FIRECRAWL_API_KEY is set, routing-and-ops.md says the page URL is sent to the Firecrawl API and a PDF costs 1 Firecrawl credit per parsed page; it also shows npm install -g firecrawl-cli. Set keys only for sources you need.
Scripts write reference files
format-converter.py writes .ris, .bib, .enw or .nbib files. Choose the output folder yourself.

Technical details

Talks to	CrossRef, PubMed (NCBI), arXiv. OpenAlex (no-MCP search script), with an optional API key. Scopus and ScienceDirect, only when requested. Semantic Scholar and Firecrawl, if you set their keys
Sign-in and keys	Elsevier keys are read from the local pybliometrics config, normally ~/.config/pybliometrics.cfg. The README says not to write them into repository files.
Scopes	Search queries, DOIs and your PubMed email Optional Elsevier access via a local pybliometrics config Reference files written to disk

Questions

What does nature-academic-search do?

It makes Claude search the literature, verify citations, build MeSH search strategies and export references. It also audits strict external citations of a target paper and profiles the influential researchers who cited it.

How do I install it?

Use the install box above to add the skill folder. That is enough for the bundled scripts, because the search server is optional. To add the server, the folder offers two routes: install.sh (pip-installs mcp-server/requirements.txt, copies the server to ~/.claude/mcp_servers, adds academic-search to ~/.claude/.mcp.json and enables it in settings.json; it also copies the skill to ~/.claude/skills/academic-search), or the uv command in static/core/routing-and-ops.md, with config/mcp-snippet.json as an example entry. Set PUBMED_EMAIL for PubMed search.

Does it work without the MCP server?

Partly. routing-and-ops.md describes a no-MCP fallback for discovery and export: scripts/academic_search.py queries OpenAlex, and scripts/format-converter.py turns DOIs, PMIDs and arXiv IDs into .ris, .bib, .enw or .nbib using CrossRef, PubMed and arXiv. The same file says the MCP path stays preferred when it is available.

Does it need API keys?

Not to start. PubMed search needs an email address (PUBMED_EMAIL). Optional keys are NCBI_API_KEY, SEMANTIC_SCHOLAR_API_KEY, OPENALEX_API_KEY, FIRECRAWL_API_KEY, and your own Elsevier access through a pybliometrics config for Scopus and ScienceDirect.

Is it safe?

It sends your search terms, DOIs and optional keys to scholarly APIs, and it can write reference files to disk. Keep Elsevier keys in the local pybliometrics config, as the README says, and not in repo files. Read the permissions section before you run the bundled install.sh.

Will it use up my Scopus quota?

Only if you ask. The MCP server searches CrossRef, PubMed and arXiv by default, and uses Scopus or ScienceDirect only when you name them or call their tools.

Can it export to Zotero, EndNote or BibTeX?

It writes .ris, .bib, .enw and .nbib files. For Zotero RDF export, the repo lists a separate nature-citation skill.

Information

updated Nov 2, 2026

Repository: yuan1z0825/nature-skills
Folder: skills/nature-academic-search

yuan1z0825/nature-skills skills.sh

Sources

Docs, issue trackers and discussions this page is built from.

Show all 7

nature-academic-search SKILL.md · github.com
README_EN.md · github.com
routing-and-ops.md · github.com
mcp-server/README.md · github.com
nature-skills README · github.com
install.sh · github.com
mcp-snippet.json · github.com