crossref-citation-check

name	crossref-citation-check
description	Validate bibliography information (authors, title, journal, volume, issue, pages, URL, DOI, and related metadata) of scholarly articles using Crossref's REST API. Use this skill when the user asks to verify or correct citations: it compares provided references against Crossref's authoritative metadata and reports discrepancies.
argument-hint	Path to a citations file (JSON or CSV) or a structured list of citation objects (JSON array). Each citation should include at least a title or DOI; other fields (authors, journal, volume, issue, pages, year, URL) are optional but recommended.
user-invokable	true

Crossref Citation Check Skill

Purpose

This skill provides a clear, repeatable workflow for verifying citation metadata against the Crossref REST API. It helps detect missing or incorrect fields (authors, title, journal, volume, issue, pages, year, URL, DOI) and suggests corrections based on Crossref's records.

When to use this skill

• The user provides a list of references and asks if they are complete or accurate.
• You need to check that a citation's DOI and other metadata match the published record.
• The user wants to identify discrepancies between their citation list and Crossref's authoritative metadata.

Steps to follow

1. Collect input: Ask the user to supply a file path or a structured list (JSON or CSV) containing citations. Each citation should include at minimum a title or DOI. Optionally include authors, journal name, volume, issue, pages, year, DOI and URL.

2. Retrieve metadata:

   • If a DOI is provided, query Crossref at https://api.crossref.org/works/<doi> to fetch authoritative metadata.
   • If no DOI is provided, search by title using https://api.crossref.org/works?query.bibliographic=<title>&rows=1 and treat the first result as the best candidate. If the top match looks unrelated, request clearer input from the user.

3. Compare fields:

   • Title: Compare the provided title to Crossref's title(s), ignoring differences in case and common punctuation. Prefer the primary title returned by Crossref.
   • Authors: Normalize author names (trim, lowercase, remove punctuation) and compare as unordered sets of full names; report missing or extra authors.
   • Journal: Compare the provided journal name to Crossref's container-title (allow common abbreviations to be flagged but not treated as exact matches).
   • Volume, Issue, Pages, Year, URL, DOI: Compare values directly with Crossref fields (volume, issue, page, published date parts, URL, DOI). For date-parts, use the first element as the year.

4. Report results:

   • For each citation, indicate whether each field matches Crossref's metadata. Clearly highlight discrepancies and any missing fields.
   • Return results in a structured format (JSON recommended) containing the provided value, Crossref value, and a match status for each field.

5. Follow up:

   • If entries match, inform the user the citation appears accurate.
   • If discrepancies exist, provide suggested corrections and the authoritative Crossref values.
   • If Crossref has no likely match, ask the user for additional details (full title, DOI, or author names) and suggest manual verification strategies.

Example input format

Here is an example of a citation list in JSON format that the user might provide:

[
  {
    "title": "Deep Learning Applications in Genomics",
    "authors": ["Alice Smith", "Bob Jones"],
    "journal": "Bioinformatics Research",
    "volume": "10",
    "issue": "2",
    "pages": "123-130",
    "year": "2025",
    "doi": "10.1000/example.doi",
    "url": "https://doi.org/10.1000/example.doi"
  },
  {
    "title": "Neural Networks for Data Analysis",
    "authors": ["Carol Lee"],
    "journal": null,
    "volume": null,
    "issue": null,
    "pages": null,
    "year": null,
    "doi": null,
    "url": null
  }
]

In the first entry, a DOI is provided, so the skill should query Crossref by DOI and compare the returned metadata. In the second entry, no DOI is provided, so the skill should search by title and compare against the closest match.

Notes

• This skill assumes the Crossref REST API is reachable. If the API is unreachable, inform the user and suggest retrying later.
• Include a polite User-Agent header with a contact email in API requests, for example: CitationChecker/1.0 (mailto:[e-mail]).
• Rate limits and polite usage: insert brief delays (for example, 1 second) between requests for bulk checks, and avoid high-frequency polling.
• When interpreting Crossref date-parts, use the first element as the year (for example, "published-print": {"date-parts": [[2024, 5, 1]]} -> year 2024).

Optional script

For convenience, this directory includes a helper script, crossref_checker.py, which automates the querying and comparison steps described above. See the script's docstring and usage examples for invocation details (for example, python crossref_checker.py -i citations.json).