CITATION.cff Metadata Builder

How to use CITATION.cff Metadata Builder

Give people who reuse your project a clear record of its name, release and credited authors. Enter facts you know, review the generated citation file and place it in your repository yourself. This local authoring form keeps person and team names separate and preserves your authored spelling.

1. Enter the project title, work type and your own citation message. Add authors in the credit order you intend, using person or entity for each CSV record.
2. Add known release metadata. Leave an optional field exactly empty to omit it; check names, dates, identifier spelling and URLs before sharing.
3. Build the file and review the complete current output. Download CITATION.cff for the repository or JSON evidence to retain every original form field and the exact generated document.
4. Place the correctly named file in your repository manually and test your chosen citation consumer. This page neither publishes a repository change nor verifies its rendering.

Good to know

• One selected CFF 1.2.0 authoring profile for software or datasets. It does not import arbitrary CFF, format APA/MLA citations, convert a reference library, infer authors, add preferred-citation/references, choose a license, resolve identifiers or edit a remote repository.
• Required title, message and 1–20 ordered authors. Person records require family names and may include given names. Entity records use a literal organization/team name. Exact duplicate author objects refuse; similar names are not merged or identified as the same real person.
• Author CSV has the exact header kind,family,given,entity, four columns and no ignored blank records. It accepts ordinary quoted CSV and one initial BOM. At most 20,000 UTF-16 units and 60,000 UTF-8 bytes. The evidence retains its original spelling and line breaks.
• Title is at most 300 UTF-16 units; message 1,000; abstract 2,000; each author name 150; version 100; each URL 2,048; DOI 300. Required and supplied optional values are nonblank. Only an exactly empty optional string is omitted. Text is never trimmed, normalized or interpreted as HTML, YAML tags or code.
• Use well-formed Unicode without C0/C1 controls, noncharacters or lone surrogates. Message and abstract additionally allow literal TAB/CR/LF and Unicode line separators; names, title, version and URLs remain on one line. YAML strings are quoted so numeric-looking names, punctuation and multiline text retain their string values.
• Release date uses a real Gregorian date in YYYY-MM-DD, years 0001–9999. URLs are credential-free canonical ASCII HTTPS spellings; the parser must reproduce the source exactly. DOI is checked against the selected CFF 1.2.0 spelling only. No URL, DOI or repository is contacted, and existence, ownership and metadata truth remain unverified.
• Both full outputs are checked before either returns: YAML at most 100,000 UTF-16 units and UTF-8 bytes; JSON at most 200,000 of each. Unsupported input refuses atomically. Copy and Download contain the complete selected current output, including personal credit information you entered.
• Download is a local file, not a repository publication or a consumer compatibility certificate. GitHub and other services apply their own rules after you manually submit the file. No account, network request, persistent history or automatic save is used.