Literature Review Synthesizer
Synthesizes multiple papers into a thematic review — where studies agree, where they conflict and why, and what nobody has studied yet.
Prompt variables
2 variables detected — fill them and Copy inserts your values (0/2 filled).
Synthesize the papers/sources below into a literature review organized by THEME, not by paper. A summary-per-paper list is an annotated bibliography, not a synthesis — I want the map of the field's agreements and fights.
**My research question:** {{research_question}}
**Sources (abstracts, notes, or full text):** {{sources}}
Structure:
**1. The landscape** (1 paragraph): what this body of work collectively tries to establish, and roughly how mature the evidence is (exploratory? converging? settled with dissenters?).
**2. Points of convergence:** findings supported by multiple independent sources. For each: the claim, which sources support it (cite by first author), and how strong the combined evidence actually is — three small studies from the same lab is NOT three independent confirmations; say so when you see it.
**3. Points of conflict:** where sources disagree. For each conflict, don't just note it — diagnose it: different populations? different measurements or definitions? different eras? methodological quality gap? If the conflict has a likely resolution ("A measured X, B measured Y, they're answering different questions"), propose it, labeled as your inference.
**4. Methodological patterns:** the field's habits and their consequences — common sample limitations, shared assumptions nobody tests, measures everyone borrows. What conclusions are hostage to these habits?
**5. The gap map:** what my research question needs that NO source provides: unstudied populations, untested mechanisms, missing longitudinal evidence. Rank the 3 gaps most relevant to my question.
**Discipline:** claims trace to specific sources; your inferences and bridging speculation are labeled [SYNTHESIS]. Where my materials are too thin to judge (abstract-only), say what the full text would need to show.
Comments (0)