The report appears here when the agents finish.
New run
Ask the literature a question. Get the evidence, the disagreements, and the gaps.
Try:
Papers
Your documents
Papers or reports you hold, private to you. Ticked ones join this run and are read and counted like any study.
Run
Countingevery shortlisted paper
A scope applies to every count from here on: agreement between papers, the exports and follow-up answers. The report already written is not rewritten.
Agent log
Papers using both:
Not combined in any paper, though both are common
Each finding a paper reports, set beside every other paper's statement about the same driver and outcome. Support is counted in independent sources: papers that share a dataset, authors or a citation count once. Agreement is not proof, and a finding only one paper reports is not unsupported.
Ask about a claim, a gap, a paper or a number. Click any id in the report or the map (C12, G2, N1) to ask about it. New counts are checked by code; the report stays as it is.
Turn this run into a first draft. A proposal argues for a gap or direction and sets out aims, methods and a workplan; a review manuscript writes the run up as a literature review with its PRISMA record. Statements about the literature are checked like the report; aims and methods are marked as your plan.
An open-ended researcher works on this run's studies on its own initiative, within a charter you set. It keeps an agenda and a notebook, and proposes patterns; code decides what counts as a finding. It sees only half of the studies: every finding must survive subgroup tests, known traps, a critic's rival explanations and confirmation on the hidden half. Its findings are its own hypotheses, graded by code, and never enter a report.