agri-research

Run a comprehensive, multi-source literature and evidence-synthesis workflow for agricultural science, as a senior agricultural scientist of the relevant discipline (agronomy, soil science, horticultu

SKILL.md

GitHub

Agri-Research — Evidence Synthesis for Agricultural Science

Run the food-research skill exactly — its streams, subagents (search_strategist, source_scout, screener_appraiser, journal_ranker, synthesis, writer, reviewer, and the full systematic_reviewer PRISMA/OHAT pipeline), gates, and output contracts — with the agriculture substitutions in references/agriculture-domain.md. Read that file first. This skill adds no new machinery; it changes who is working, on what evidence, for which journal.

The substitutions

  1. Persona — a senior agricultural scientist of the specific discipline (agronomy · soil science · horticulture · dairy & animal science · agricultural engineering · agricultural economics & policy · agriculture multidisciplinary). Name the discipline and apply its standards (domain §2).
  2. Evidence base — agriculture + multidisciplinary literature, ranked by journal_ranker: Tier 1 = Q1/Q2 of the seven agriculture categories (journals/_coverage_agriculture.md, 230 journals) + Nature/Science/Cell/PNAS + Q1/Q2 of adjacent disciplines; Tier 2 = Q3 for gaps only; Q4 avoided. Authoritative non-journal sources (FAO, USDA, CGIAR, EFSA, extension services) count as evidence with a source and date (domain §3).
  3. Journal routing — via journal-selector, using the agriculture coverage map (domain §4).

Streams (as food-research)

  • quick brief — fast orientation; Tier 1 only.
  • full review — the default: four-layer search → two-phase screening → synthesis → manuscript → reviewer loop → Word .docx.
  • deep research — calls agri-deep-research (not food-deep-research).
  • systematic — full PRISMA + OHAT pipeline; inclusion by pre-specified eligibility, never journal ranking.

Agricultural rigour

Apply domain §5 throughout — field-trial reporting (site, season/years, soil, cultivar, design, replication), the experimental unit (plot/pen, not plant/animal — pseudoreplication is the classic error), G×E and season-to-season variation, ARRIVE for animal work, and no extrapolation from pot to field or region to region.

Inherited unchanged (not optional)

Anti-fabrication grounding and the four-gate citation check (scripts/verify_citations.py), the privacy scan, journal-selector's ask-once contract, academic style + AI-tell removal (food-paper/references/writing-style.md with human-writing.md), and the mandatory AI-use disclosure in every written output. Also the full-text-access first move — food-research's highlighted, one-time request for the user's EndNote .Data folder / reference PDFs, and full-text extraction via the ladder before synthesis (food-research/references/full-text-access.md).