Nature/CNS Citation Skill
Add strict Nature/CNS citations to your manuscript by splitting text into citable segments, searching only Nature Science Cell journals, and exporting reference-manager files.
The skill's instructions show it only queries Crossref and PubMed APIs with no data exfiltration or excessive permissions. It requests a user email for polite pool compliance, but lacks an explicit user confirmation step before sending requests. There is author metadata integrity checking to prevent exports of incomplete records, but details on rate-limit handling, failure recovery, and rollback are incomplete. Therefore, not full marks; 2 points deducted.
Static review only; no execution. The script has retry mechanisms (--max-retries) and batching, but there is no dedicated test suite for this script in the repository CI (CI covers other skills). Error handling for abnormal inputs (e.g., blocking exports with missing authors) is described, but no concrete failure-message examples are provided. Thus, limited to 7 per static review cap.
The skill clearly defines scenarios (manuscript passages, claims, reviewer responses) and boundaries (candidates, not guarantee). Trigger semantics are rich with Chinese and English keywords, aiding precise invocation. Environment fit: supports Chinese users, but core function depends on overseas services (Crossref, PubMed) which may be partially inaccessible from mainland China, though not entirely. Score 12.
Documentation is well-structured with README, README_EN, and detailed reference files, following progressive disclosure. Version 2.1.0 and Apache-2.0 license are present. Maintenance responsibility is clear (by Yuan1z). However, no changelog, versioning only in manifest, and skill marked Beta, but overall governance good. Deduct 3, score 12.
Static review cannot confirm actual output usability. By design, it completes the core task (segmentation, search, export citation files) and provides inspectable RIS files with author-integrity checks, reducing error risk. But without execution, outputs need user review. Therefore, limited to 7 per static cap.
Repository CI covers other skills, not this one, so no automated tests for this skill. However, evaluation cases exist to demonstrate expected behavior, and reference files detail implementation, enabling manual verification. Evidence source is single (repository itself). Score 5.
- The skill depends on Crossref and PubMed APIs which may be inaccessible from mainland China without a proxy; prepare alternatives.
- Due to static review limitations, the script was not executed; test on a small sample before real use.
- Using '--allow-incomplete-authors' bypasses author integrity checks and may result in incomplete metadata; use with caution.
- Exported files require user review; the skill does not guarantee citation suitability; verify accuracy.
What does this skill do, and when should you use it?
nature-citation is part of the nature-skills collection, designed to add rigorous Nature/CNS citations to academic manuscripts. It splits long passages into citable segments, searches only within Nature Portfolio, AAAS Science, and Cell Press flagship and subjournals, filters by publication time, and exports a single reference-manager-ready file (RIS, EndNote, or Zotero RDF). The skill uses a router architecture, loading core principles and workflow from static fragments, and invokes a Python script for search and export.
Reads user-supplied text and splits it into citable segments. Builds search queries restricted to Nature, Science, and Cell journals, with scope set by user wording (flagship-only versus subjournals). Searches online databases for candidate papers and evaluates their support for each statement, avoiding title-only or metadata-only matches. Validates complete author metadata by refetching via PMID if given names are missing. Exports one reference-manager file (RIS, EndNote, or Zotero RDF) by default. Relies on Python script scripts/nature_citation.py and network access.
- A PhD student wants to add citations to an Introduction draft with EndNote compatibility.
- A researcher needs to replace weak references with strict Nature/Science/Cell papers before submission.
- A non-native English speaker writes prompts in Chinese but wants English search queries.
- An author filters citations to the last five years to emphasize recent advances.
- A user needs a Zotero RDF file to import references directly into Zotero.
What are this skill's strengths and limitations?
- Strict journal scope ensures high-impact support from Nature, Science, Cell.
- Automatic text segmentation provides clear statement-to-reference correspondence.
- Exports standard reference formats: RIS, EndNote, Zotero RDF.
- Router design keeps core small and loads heavy references only when needed.
- Conservative support evaluation prevents false or weak citations.
- Only covers Nature/Science/Cell family, limiting scope for topics needing broader literature.
- Requires network and external script; Python dependencies may not be pre-installed.
- Marked as Beta, so edge cases are possible and no test suite is publicly provided.
- Installation instructions focus on Codex; Claude Code requires manual setup.
How do you install this skill?
Install the entire nature-skills repository. Using Codex: npx skills add Yuan1z0825/nature-skills --skill nature-citation --skill nature-shared --yes --copy. For Claude Code, clone the repo and copy the skill directories to ~/.claude/skills/ or use a subagent wrapper. Ensure Python and any required dependencies are installed (specific requirements not documented in SKILL.md).
How do you use this skill?
Trigger with natural language requests for adding citations (e.g., 'add Nature-style references to this paragraph'). The skill auto-detects scope (Nature, CNS, subjournals) and language. Ensure network access for searching. For long texts (>10 segments), the skill will use a batched strategy.
How does this skill compare with similar options?
Compared to other skills like nature-academic-search (which offers broader literature search) or nature-ref-verifier (which validates citations), this skill specializes in adding citations with strict journal restrictions.