Nature Citation

{
  "skill_name": "nature-citation",
  "evals": [
    {
      "id": 1,
      "prompt": "把这段文字自动分段并给出Nature/CNS及其子刊引用，导出Zotero RDF格式和HTML可视化：Tumor-associated macrophages promote immune evasion by suppressing cytotoxic T cell activity. Single-cell RNA sequencing reveals cellular heterogeneity in pancreatic cancer.",
      "expected_output": "Segments the text, maps each segment to citation candidates, exports references.rdf, and provides an HTML visualization that can download selected references as ENW, RIS, or Zotero RDF.",
      "files": []
    },
    {
      "id": 2,
      "prompt": "只看Nature系列，把下面这一段按长度分段，给我文本和引用的对应关系，方便插入论文：single-cell RNA sequencing reveals cellular heterogeneity in pancreatic cancer. Spatial transcriptomics further preserves tissue context for interpreting tumor microenvironments.",
      "expected_output": "Restricts scope to Nature Portfolio-style journals and produces a segment-reference correspondence table.",
      "files": []
    },
    {
      "id": 3,
      "prompt": "Find flagship Nature/Science/Cell references for: CRISPR screens can identify genetic dependencies in cancer cells. Export RIS for EndNote.",
      "expected_output": "Restricts to Nature, Science, and Cell only, treats the claim as a segment, and exports RIS without fabricating missing metadata.",
      "files": []
    },
    {
      "id": 4,
      "prompt": "给我一个Nature系列引用导出，用户要自己选择下载 ENW、RIS 还是 Zotero RDF，并且先按年份筛选再勾选参考文献。",
      "expected_output": "Produces the citation browser HTML with year filters, selectable references, and downloadable ENW/RIS/Zotero RDF exports from the same page.",
      "files": []
    }
  ]
}


name: nature-citation
version: 2.0.0
description: >
  Declarative manifest for the static/dynamic split. SKILL.md uses this to
  decide which fragments to load for a citation request.

# Design note: nature-citation is a linear, parameterised workflow (segment ->
# parse -> search -> evaluate -> export -> report). Its variation — journal
# scope, user language, input length — is handled at runtime by inline rules and
# by the nature_citation.py flags (--scope, --batch-size, ...), not by loading
# different large content fragments. There is therefore no content axis; the
# split is core (always loaded) plus on-demand references. Heavy, only-sometimes-
# needed material (the script flag reference and long-article batching strategy)
# lives in references so a normal short run stays cheap. nature-citation does not
# use the prose-oriented _shared layer.

always_load:
  - static/core/principles.md
  - static/core/chinese-mode.md
  - static/core/workflow.md

references:
  on_demand:
    - condition: running nature_citation.py — full flag list, long-article batch strategy, quick-guide tables
      path: references/script-usage.md
    - condition: translating a manuscript claim into search queries and support grades
      path: references/search-strategy.md
    - condition: the default Nature/CNS journal-family boundary and official source notes
      path: references/journal-scope.md
    - condition: RIS, EndNote, or Zotero RDF export guidance
      path: references/ris-endnote.md
    - condition: segment text, search Crossref, export ENW/RIS/RDF, and generate the HTML browser
      path: scripts/nature_citation.py


# `nature-citation` skill

A citation-search skill for turning manuscript text or standalone claims into strict Nature / CNS-family reference exports with segment-level mapping and reference-manager-ready downloads.

This skill is bilingual-aware. It accepts Chinese manuscript text and citation requests such as "分段引用", "Nature系列引用", "CNS及子刊", "补引用", "支撑文献", or "导出 Zotero", then searches with English scientific concepts while returning Chinese review notes by default.

## What it does

- splits manuscript text into citable segments with stable IDs such as `S001`, `S002`, and `S003`
- converts each segment into search queries for Crossref-led discovery
- filters results to Nature Portfolio, the AAAS Science f