# Search academic literature across five databases

Academic researchers struggle to search multiple databases, verify citations, and manage reference files separately. This skill unifies PubMed, CrossRef, arXiv, Scopus, and ScienceDirect into coordinated workflows with citation conversion and deduplication.

## Install

```bash
npx skillstore add community contribution, refactored into static/dynamic layers/yuan1z0825-nature-academic-search
```

## Metadata

- - Slug: yuan1z0825-nature-academic-search
- - Version: 2.0.0
- - Author: Community contribution, refactored into static/dynamic layers
- - GitHub username: Yuan1z0825
- - License: MIT
- - Repository: https://github.com/Yuan1z0825/nature-skills/tree/main/skills/nature-academic-search
- - Ref: main
- - Supported tools: Claude, Codex, Claude Code
- - Risk level: low
- - Risk factors: external\_commands, filesystem, env\_access, scripts, network
- - Quality score: 82
- - Quality tier: silver
- - Public page: https://skillstore.pages.dev/skills/yuan1z0825-nature-academic-search
- - Manifest: https://skillstore.pages.dev/api/skills/yuan1z0825-nature-academic-search/manifest

## Capabilities

- Multi-source concurrent search across PubMed, CrossRef, arXiv, Scopus, and ScienceDirect
- Automatic identifier detection for DOI, PMID, and arXiv IDs
- Citation formatting in APA, Nature, IEEE, and Vancouver styles
- MeSH vocabulary lookup for building PubMed search strategies
- Citation file conversion between .nbib, .ris, .bib, and .enw formats
- Duplicate detection across sources using DOI/PMID hashing

## Use Cases

- Literature review for a research paper: Search multiple databases simultaneously for a topic, deduplicate results, and export citations in the required format.
- Citation verification and bibliography cleanup: Verify citations extracted from a manuscript against authoritative sources and fix formatting errors.
- MeSH search strategy construction: Build a precise PubMed search strategy using MeSH vocabulary lookup and Boolean operators.

## Prompt Templates

### Basic multi-source search

```
Search for recent papers about CRISPR gene editing in cancer therapy across PubMed, CrossRef, and arXiv. Return the top 10 results with abstracts.
```

### Citation verification workflow

```
I have a list of citations from a manuscript. Verify each one against PubMed and CrossRef, flag any that cannot be confirmed, and return corrected entries.
```

### MeSH strategy building

```
Build a PubMed MeSH search strategy for a systematic review on 'machine learning in radiology diagnosis'. Include MeSH terms, synonyms, and Boolean structure.
```

### Citation file format conversion

```
Convert my .nbib file from PubMed into BibTeX format. Also export to .ris for EndNote compatibility.
```

## Limitations

- Google Scholar and Semantic Scholar support is limited; results may vary
- Chinese literature databases \(CNKI, Wanfang\) are not indexed
- Citation counts may be delayed; CrossRef updates monthly
- Scopus and ScienceDirect require Elsevier API entitlement and consume quota

## Best Practices

- Start with the T1 sources \(PubMed for biomedical, CrossRef for cross-disciplinary, arXiv for preprints\) before falling back to T2/T3
- Always provide a contact email via PUBMED\_EMAIL env var or config.toml to comply with NCBI rate limit policies
- Run scripts/preflight.py before batch operations to verify API endpoints are reachable
- Use the dedicated Scopus/ScienceDirect tools only when needed, as they consume Elsevier API quota

## Anti Patterns

- Do not hardcode API keys in plugin files; use environment variables or config.toml instead
- Do not run the full MCP server for single-source queries; use the direct source modules for efficiency
- Do not skip deduplication when merging results from multiple sources; DOIs and PMIDs often overlap

## Security Audit

- - Safe to publish: true
- - Audited at: 2026-06-24T05:54:22.914\+00:00
- - Summary: Legitimate academic literature search skill with MCP server integration for PubMed, CrossRef, arXiv, Scopus, and ScienceDirect. Static analyzer flagged 692 patterns, but the vast majority are false positives: markdown inline code formatting \(backticks\), MD5 hashes used for citation deduplication \(not security cryptography\), references to hidden config files \(.mcp.json, .config/pybliometrics.cfg\), and legitimate network calls to public academic APIs. No confirmed malicious patterns detected.

## Stats

- - Views: 0
- - Downloads: 13
- - Favorites: 1
- - Popularity score: 0