--- name: literature-search-strategy description: Design, adapt, document, or execute reproducible scholarly literature searches for a defined research question. --- # Literature search strategy Use for search design or documented execution, including updating an existing strategy. First establish the question, purpose, date range, languages, relevant study types, and whether sensitivity or precision is the priority. If a decision cannot be inferred, state a reasonable provisional choice. ## Construct and check Translate the question into essential concepts. For each, assemble synonyms, spelling variants, abbreviations, and relevant controlled vocabulary; combine synonyms with OR and concepts with AND. Avoid forcing every question element into the search when it would exclude relevant work. Apply exclusions and filters cautiously and explain their anticipated effect. Select sources based on subject coverage and access. Translate syntax separately for each interface: field tags, phrase searching, truncation, proximity, vocabulary expansion and limits vary. Verify current provider documentation when exact syntax matters. Check the strategy against a small set of known relevant records if available, investigate misses, and record revisions; such a check does not prove completeness. When authorized and tools permit, run searches and preserve each source, platform, full query, date, limits, record count, and export format. Record failures, rate limits, and inaccessible sources. Deduplicate with a traceable record of rules and potential uncertain matches; do not remove reports solely because they share a study. ## Output and dependencies Provide ready-to-use source-specific queries and a search log, or label them as drafts if not run. List planned supplementary searches, including reference or citation checks when useful. Search interfaces and access are external dependencies; never claim a search ran, records were retrieved, or coverage is comprehensive without evidence. Cite actual sources and dates for performed searches and clearly distinguish incomplete coverage from a negative result.