Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →brycewang-stanford/anystyle-api
Citation reference parser using machine learning
Passed 22 of 23 test cases with the skill, and 15 of 23 without it.
and 6 more in the full report
7 fixed15 passed either way1 still failing0 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 7 test cases the skill fixed. Same model, same prompt, with and without the skill.
I want to test local reference parsing on a single citation string directly from my terminal using the CLI tool without writing a text file first. What is the syntax for parsing a single string argument?
| Checked | Without skill | With skill |
|---|---|---|
| Passes the reference string argument directly to anystyle parse. | Missed | Passed |
The instructions your model reads when the skill loads.
AnyStyle is a fast and smart citation reference parser that uses machine learning (specifically conditional random fields, CRFs) to extract structured bibliographic data from unformatted citation strings. It can parse raw reference text into structured fields such as author, title, journal, volume, pages, year, and DOI, handling the enormous variety of citation formats found in academic literature.
The AnyStyle service provides both a web interface and an API endpoint for programmatic citation parsing. Unlike rule-based parsers that rely on specific citation style templates, AnyStyle uses a trained machine learning model that generalizes across citation formats, making it effective for parsing references from diverse disciplines and publication traditions where citation styles vary widely.
Researchers, librarians, digital humanists, and research software developers use AnyStyle to extract structured references from PDF documents, legacy bibliographies, dissertation reference lists, and scanned documents. It is particularly valuable for building citation networks, enriching bibliographic databases, migrating references between management tools, and processing large volumes of unstructured citation data that would be impractical to parse manually.
No authentication required. The AnyStyle web service is freely accessible without any API key, token, or registration. The service can be used via the web interface at https://anystyle.io/ or through its API endpoint. For heavy usage or private deployments, AnyStyle is also available as an open-source Ruby gem that can be installed locally.
Submit raw citation text and receive structured bibliographic data extracted by the machine learning parser. The endpoint accepts one or more citation strings and returns parsed fields for each reference.
POST https://anystyle.io/parse| Parameter | Type | Required | Description |
|---|---|---|---|
| body | string | Yes | Raw citation text (one reference per line in the POST body) |
| format | string | No | Output format: json (default), xml, bib (BibTeX) |
# Parse a single citation
curl -X POST "https://anystyle.io/parse" \
-H "Content-Type: text/plain" \
-d "Vaswani, A., Shazeer, N., Parmar, N., et al. (2017). Attention is all you need. Advances in Neural Information Processing Systems, 30, 5998-6008."
# Parse multiple citations (one per line)
curl -X POST "https://anystyle.io/parse" \
-H "Content-Type: text/plain" \
-d "Vaswani, A. et al. (2017). Attention is all you need. NeurIPS 30, 5998-6008.
test cases passed, vs 15 / 23 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.