Examples by paper type

Start with the question, material and stage of your paper. These routes show how the shared Skills can fit different studies; your agreed protocol determines which steps are required. A paper-type label does not require every stage to run.

Systematic review (systematic-review)

Ask: “Help me carry out this review under the supplied protocol. Start by checking the search strategy and screening criteria.”

Clarify the question, design the search, screen and extract studies, assess their quality, then synthesise the evidence. Choose reporting checks appropriate to the review. Retain search and screening logs, extraction tables, exclusion reasons, quality assessments and the manuscript's evidence links. An existing corpus lets you resume at extraction or synthesis only when the preceding decisions are available. See rigorous literature search.

Empirical study (empirical)

Ask: “Review the design of this policy-effects study using the available data and our agreed analysis plan.”

Use question and design guidance from Stages A/C, research code from Stage I when needed, and Stage F to write from actual results. Check measurement, identification and robustness according to the method. A completed study can start with writing and reporting checks while preserving its design and analysis records.

Qualitative study (qualitative)

Ask: “Help turn these authorized interview excerpts and analytic memos into a findings section. Keep source locations, rival interpretations and limits visible.”

Let the method guide sampling, coding, comparison and interpretation. Use relevant design, ethics, writing and reporting guidance; do not impose hypotheses or statistical robustness checks on every qualitative study. Retain the sampling rationale, protocol, memos, evidence trail and applicable reporting checklist.

Methods paper (methods)

Ask: “Review this proposed method and evaluation design, then prepare the code specification and implementation plan.”

Stage I connects specification (I5), planning (I6), implementation (I7), review (I8) and reproducibility (I4). State comparison methods, data, metrics and limits before claiming an improvement. The code, evaluation outputs and reproducibility record are part of the evidence. Preserve those links in the manuscript.

Theory paper (theory)

Ask: “Help clarify the mechanism in this argument and test it against the supplied theoretical sources and competing explanations.”

Use framing, theory mapping and synthesis to develop the argument, then write and review its logic. Keep propositions, assumptions and theoretical claims distinct from empirical evidence. Add code or simulation only when it contributes to the research question.

A smaller follow-up

Name the specific section, decision or finding you want changed and the material that should support it. The model can choose a short route while preserving required evidence and write approvals. For an independent critique, request an actual subagent or cross-Host review.

See task examples, the Skills map and the complete Graph example.