Literature research

Start#

/research Does heart rate variability (HRV) biofeedback reduce perceived stress in adults? Focus on randomized controlled trials and report effect sizes.

A good question names the population, the intervention or variable, and what you want emphasized (study designs, outcomes, time range).

What happens#

1. Clarifying survey — databases, years, what to emphasize. Skip it to use defaults (all databases, no date limit).
2. Search — parallel queries against OpenAlex, PubMed, Crossref, Semantic Scholar, Europe PMC and the web, with keyword tracks derived from your question.
3. Screening — each candidate is kept or dropped against your question; the decisions are saved as a table you can audit.
4. Synthesis — a report with numbered citations, plus a citation graph and a research mind map.
5. Review — a reviewer checks that claims are supported by the cited sources before the turn ends.

What you get#

• <topic>_report.md — background, research landscape, key findings with effect sizes when reported, milestones, open questions, references.
• literature_screening.csv — every candidate with the include/exclude decision and reason.
• <topic>_citation_graph.svg and <topic>_research_mindmap.svg.

Steer it#

• Narrow or widen: "restrict to 2018–2026", "include grey literature", "only RCTs".
• Ask follow-ups in the same session: "which of these trials measured HRV coherence?" — the agent uses the screened corpus first.
• Effort: Lite gives a sourced chat answer without a report (cheapest); Basic writes a 3–4 section report; Deep searches without a cap and writes a longer report. See Effort levels & models.
• Patents — /patents <topic> runs the same pipeline against patent databases and adds a filing-trend chart.

Walkthrough with screenshots: Literature review, step by step.