Literature Review

Literature Review

Use this skill to turn a broad research question into a traceable review process with explicit search logic, screened evidence, and defensible synthesis.

Workflow

1. Define the review question.

   • Convert vague topics into a focused question using PICO, SPIDER, PCC, or a simpler topic-scope-outcome frame.
   • State population/domain, intervention or concept, comparator if any, outcomes, timeframe, geography, and document types.
   • If the user has not provided scope, make conservative assumptions and label them.

2. Plan the search.

   • Use references/search-strategy.md when designing queries, databases, source types, or grey literature searches.
   • Build query blocks from concepts, synonyms, acronyms, variant spellings, standards/regulations, and product or method names.
   • For current, medical, legal, regulatory, or safety-sensitive topics, browse and cite primary or authoritative sources.

3. Screen sources.

   • Use references/screening-and-extraction.md for inclusion/exclusion criteria and extraction fields.
   • Prefer peer-reviewed papers, guidelines, standards, systematic reviews, official regulatory sources, and primary datasets.
   • Exclude weak or irrelevant sources explicitly when they could otherwise bias the synthesis.

4. Extract evidence.

   • Capture citation, year, jurisdiction/context, study type, sample/data, method, outcomes, limitations, and relevance.
   • Preserve URLs, DOI/PMID/arXiv IDs, publication dates, and access dates when available.
   • Do not overstate claims from abstracts alone; flag when only an abstract or secondary summary was available.

5. Assess quality and bias.

   • Use references/evidence-synthesis.md for study-quality prompts and synthesis structure.
   • Match appraisal to source type: randomized study, observational study, qualitative study, guideline, standard, review, dataset, or technical report.
   • Separate evidence strength from volume of publications.

6. Synthesize and write.

   • Organize by themes, chronology, methods, mechanisms, jurisdictions, outcomes, or evidence strength.
   • Highlight consensus, disagreement, gaps, uncertainty, and practical implications.
   • Include a source table for substantial reviews unless the user asks for prose only.

Output Patterns

For quick reviews, provide:

• Search scope and assumptions
• Key findings
• Evidence table
• Gaps and caveats
• Sources

For formal reviews, provide:

• Research question
• Methods and search strategy
• Inclusion/exclusion criteria
• Screening summary
• Evidence extraction table
• Thematic or quantitative synthesis
• Limitations
• References

For related-work sections, provide polished prose plus a short table showing how each cited work supports the narrative.

Quality Rules

• Cite sources for factual claims about papers, guidance, standards, regulations, current evidence, or market/clinical state.
• Prefer primary sources over blogs and tertiary summaries.
• Distinguish "no evidence found" from "evidence of no effect."
• Do not invent citations, DOIs, study details, sample sizes, or findings.
• Mark low-confidence claims and explain what would verify them.
• When the topic is medical, regulatory, legal, or financial, keep recommendations informational and source-grounded.

References

• references/search-strategy.md: Query construction, source selection, and search logging.
• references/screening-and-extraction.md: Criteria, extraction tables, and evidence matrix fields.
• references/evidence-synthesis.md: Quality appraisal, synthesis structures, and final review checklist.