Extract Citations

Overview

LEVEL 1 (syntactic extraction) ONLY: finds citation-like tokens — URLs, DOIs, Markdown footnotes, Markdown links, numeric references, and author-year mentions — using bounded, ReDoS-safe pattern heuristics. It does NOT do LEVEL 2 bibliographic parsing: it never splits a reference into author/title/year/journal fields, never validates a DOI resolves, and never fetches anything. Best-effort: returns candidates with offsets and a heuristic confidence.

You often need to spot references before summarizing or auditing content. We extract candidates deterministically instead of relying on fragile per-prompt regex.

When to use it

Scope

Non-goals

• Does not fetch or validate URLs
• Does not resolve DOIs or build bibliographies
• Does not parse references into structured author/title/journal fields

Properties

Capabilities

Cost & billing

Flat rate per invocation

File handling

How we handle your data

Input schema

Output schema

API contract

curl -sS -X POST "$BASE/v1/execute" -H "Authorization: Bearer $KEY" -H 'Content-Type: application/json' -d '{"skillKey":"text.extract_citations","input":{"text":"See https://example.com"}}'

import { Squillify } from "@squillify/sdk";

const sq = new Squillify({ apiKey: process.env.SQUILLIFY_API_KEY });

const { output } = await sq.execute("text.extract_citations", {"text":"See https://example.com and doi:10.1000/xyz123"});

Errors

Other limits

Limitations

• Does not fetch or validate URLs
• Does not resolve DOIs or build bibliographies
• Does not parse references into structured author/title/journal fields
• Authoritative bibliography parsing with a formal grammar
• Verified reference validation or DOI resolution
• Fetching linked content from extracted URLs

How to call

POST /v1/execute (requires authentication)

{
  "skillKey": "text.extract_citations",
  "input": {
    "text": "See https://example.com and doi:10.1000/xyz123"
  }
}

Examples

{
  "text": "See https://example.com and doi:10.1000/xyz123"
}

{
  "truncated": false,
  "citations": [
    {
      "type": "doi",
      "start": 32,
      "confidence": 0.98,
      "end": 46,
      "value": "10.1000/xyz123",
      "normalized": "10.1000/xyz123"
    },
    {
      "type": "url",
      "start": 4,
      "confidence": 0.95,
      "end": 23,
      "value": "https://example.com",
      "normalized": "https://example.com"
    }
  ],
  "ok": true
}

{
  "types": [
    "numeric_reference",
    "author_year"
  ],
  "text": "See [1] and (Smith, 2020) for details."
}

{
  "truncated": false,
  "citations": [
    {
      "type": "numeric_reference",
      "start": 4,
      "confidence": 0.6,
      "end": 7,
      "value": "[1]",
      "normalized": "[1]"
    },
    {
      "type": "author_year",
      "start": 12,
      "confidence": 0.5,
      "end": 25,
      "value": "(Smith, 2020)",
      "normalized": "(Smith, 2020)"
    }
  ],
  "ok": true
}

Recommended use cases

• Discover references before summarization or compliance review
• Surface DOIs and URLs in research notes or drafts
• Pre-filter documents for downstream citation processing

Avoid for

• Authoritative bibliography parsing with a formal grammar
• Verified reference validation or DOI resolution
• Fetching linked content from extracted URLs

Related skills

Composable with

Safety & privacy

• Runs without calling any external service.
• Your payloads are never stored permanently.
• Telemetry captures metadata only — never your payload contents.
• We never train models on your payloads.
• No human reviews your payloads.

Tags