Literature Review

• Conducting a systematic literature review, scoping review, or meta-analysis for a research paper, thesis, or grant application.
• Synthesizing the current state of knowledge on a biomedical, scientific, or technical topic across multiple databases and publication types.
• Writing the literature review section of a manuscript and needing verified citations formatted in APA, Nature, Vancouver, Chicago, or IEEE style.
• Identifying research gaps, seminal papers, and future directions in a domain by searching across preprint servers and peer-reviewed literature simultaneously.

• You only need a quick background summary on a topic rather than a documented, reproducible systematic search — a general web search skill may be sufficient.
• Your domain is outside biomedical, scientific, or technical fields, as the specialized database integrations (PubMed, bioRxiv, arXiv, etc.) will provide limited benefit.
• You do not need citation verification or PDF output and simply want informal notes on a subject.

Searches across PubMed, bioRxiv, medRxiv, arXiv, Semantic Scholar, and more via integrated scientific skills (gget, bioservices, datacommons-client). Broad academic web coverage is provided by the parallel-web skill (parallel-cli search) with scholarly domain filtering, ensuring no major source is missed.

All citations are verified for accuracy using scripts/verify_citations.py, which extracts DOIs, resolves them via CrossRef, and retrieves authoritative metadata. Supports multiple citation styles including APA (7th edition), Nature, Vancouver, Chicago, and IEEE.

Converts completed reviews from markdown to professionally formatted PDFs via scripts/generate_pdf.py, powered by pandoc and XeLaTeX. Options include selectable citation styles, table of contents, and section numbering.

Guides the review through Planning & Scoping, Systematic Search, Screening & Selection, Data Extraction & Quality Assessment, Thematic Synthesis, Citation Verification, and Document Generation — following PRISMA and Cochrane best practices.

scripts/search_databases.py combines results from all databases, removes duplicates by DOI or title, ranks by citation count, and exports clean aggregated results in markdown or JSON for screening.

Integrates with the scientific-schematics skill to produce publication-quality figures including PRISMA flow diagrams, thematic synthesis diagrams, and conceptual frameworks — colorblind-friendly and high-contrast by default.

A researcher studying CRISPR gene therapy for sickle cell disease uses the skill to search PubMed, bioRxiv, and Semantic Scholar with MeSH terms and Boolean operators, screen hundreds of results against PICO-defined inclusion criteria, extract data, assess bias with the Cochrane Risk of Bias tool, and produce a PRISMA-compliant PDF with Nature-style citations.

A graduate student writing a thesis uses the skill to conduct a multi-database search, organize findings thematically rather than study-by-study, verify all citations, and export a formatted markdown and PDF chapter ready for submission — complete with a documented search methodology for reproducibility.

A technical researcher surveys machine learning literature by querying arXiv (cs.LG, stat.ML categories), Semantic Scholar, and general academic sources via parallel-web, prioritizing highly-cited papers from top venues (NeurIPS, ICML) and synthesizing trends, benchmarks, and open research questions.

A team preparing a grant application uses the skill to quickly scope a research area across multiple databases, identify key seminal works and research gaps, and generate a concise, citation-verified summary document that supports the scientific rationale section of the proposal.