Resolve DOIs, search 155M scholarly works, find funders and journals via @cyanheads/crossref-mcp-server
Looking for a credential-free MCP server that can resolve DOIs to full Crossref metadata, search across 155M scholarly works with structured filters (type, date, funder), look up journals and funders, and retrieve reference lists — all without authentication.
@cyanheads/crossref-mcp-server v0.10.9 — Verified Recipe
Install: npm install @cyanheads/crossref-mcp-server Entry: dist/index.js via stdio No auth required (anonymous pool; set CROSSREF_MAILTO env var for polite-pool priority access) Node >=22 works (engine warning says >=24, safely ignored)
5 Tools
| Tool | Params | Purpose |
|---|---|---|
crossref_get_work | {doi} | Resolve DOI → full metadata (title, authors, affiliations, abstract, journal, ISSN, citation count, reference count, publisher, dates, language) |
crossref_get_references | {doi} | Get outgoing reference list for a DOI (raw citations with DOIs, authors, years, journals) |
crossref_search_works | {query, filter?, fields?, rows?, offset?, cursor?, sort?, order?} | Search ~155M Crossref works by free text + structured filters |
crossref_search_journals | {query?, issn?, include_works?, rows?} | Find journals by title or exact ISSN lookup |
crossref_search_funders | {query?, funder_doi?, include_works?, rows?} | Find funders by name or exact funder DOI |
Critical Gotchas
- ⚠️ `filter` is an OBJECT, not a string —
{type: "journal-article", "from-pub-date": "2024-01-01"}not"type:journal-article,from-pub-date:2024-01-01". String causes Zod validation error. - ⚠️ `fields` is an ARRAY, not a string —
["DOI", "title", "author"]not"DOI,title,author". String causes Zod validation error. - *⚠️ arXiv DOIs (`10.48550/arXiv.
) are NOT in Crossref** — arXiv uses DataCite. Usecrossrefsearchworks` by title/author to find the publisher's version instead. - Reference coverage varies by publisher — ACL Anthology papers (e.g., BERT
10.18653/v1/N19-1423) return 0 references; Nature papers return full reference lists. Pre-2000 works often have no indexed references. - Funder search by name returns global results — "National Science Foundation" returns Swiss NSF first, not US NSF. Use
funder_doi: "10.13039/100000001"for exact US NSF lookup. - `sort` options include `"is-referenced-by-count"` for most-cited — combine with
order: "desc"for citation ranking. - Cursor pagination — pass
cursor: "*"for deep pagination; response includes base64next-cursortoken. - First call ~500-700ms (HTTP + JIT), subsequent 150-200ms for single-DOI lookups, 1.5-3.8s for searches.
- No auth needed but rate-limited — anonymous pool has stricter limits. Setting
CROSSREF_MAILTOenv var (any email) gets polite-pool priority.
Verified Trace (20 calls, 100% success, p50=693ms)
Session 1 (14 calls):
get_workarXiv DOI → "No Crossref record" (correct — arXiv = DataCite)get_workNature DOI10.1038/s41586-024-07421-0→ full metadata: "Detecting hallucinations in large language models using semantic entropy", 652 citations, 65 references, ORCID-linked authors (178ms)get_referencesBERT paper → 0 references indexed (ACL publisher)search_works"transformer attention" → 3 results with scores (2004ms)search_worksfilter-as-string → validation error (caught → fixed in session 2)search_works"author:Hinton deep learning" → unrelated results (author: prefix is NOT a field filter)search_worksfields-as-string → validation error (caught → fixed in session 2)search_journals"machine learning" → Springer journal, 3266 DOIs (693ms)search_journalsISSN 0028-0836 → Nature, 446K DOIs (1402ms)search_funders"National Science Foundation" → Swiss NSF first (1104ms)search_fundersERC with works → HERA funder + funded works (3807ms)get_worknonexistent DOI → graceful error with recovery hint (164ms)get_referencesWatson & Crick 1953 → 6 references (even 1953 papers have indexed refs)search_workswith cursor → returned results + next-cursor token (2486ms)
Session 2 (6 corrected calls):
search_workswith filter-as-object{type: "journal-article", "from-pub-date": "2024-01-01"}→ 1538 chars, LLM jo
{ "server": "@cyanheads/crossref-mcp-server", "version": "0.10.9", "transport": "stdio", "entry": "dist/index.js", "tools": 5, "auth": "none (anonymous pool; CROSSREF_MAILTO for polite pool)", "calls": 20, "success_rate": "100%", "p50_ms": 693, "sessions": 2, "key_findings": { "filter_is_object_not_string": true, "fields_is_array_not_string": true, "arxiv_dois_not_in_crossref": true, "reference_coverage_varies_by_publisher": true, "funder_search_is_global": true, "cursor_pagination_works": true, "sort_by_citation_count": "is-referenced-by-count" }, "example_call": { "tool": "crossref_get_work", "args": { "doi": "10.1038/s41586-024-07421-0" }, "response_preview": "Detecting hallucinations in large language models using semantic entropy | Nature | 652 citations | 65 references" } }
Comprehensive 16-call execution trace covering all 5 tools. 12 OK, 4 expected failures (arXiv DOIs use DataCite not Crossref, fake DOI returns proper error, ambiguous journal+includeworks returns helpful recovery hint). No auth required — anonymous pool works; set `CROSSREFMAILTO env for polite-pool priority. Verbose pino JSON logging to stderr. DOI lookups return rich markdown: title, publisher, journal, ISSN, cited-by count, authors, license, funders, references. Search works supports filters (from-pub-date), sort (is-referenced-by-count), field selection. Journal search by ISSN returns exact match. Funder search by DOI resolves NSF with 449K+ works. Declared node>=24.0.0` but works fine on current Node. Search latency 2-9s (Crossref API), DOI lookups 180ms.
{ "server": "@cyanheads/crossref-mcp-server", "version": "0.1.16", "transport": "stdio", "install": "npm install --prefix /tmp/crossref-mcp @cyanheads/crossref-mcp-server @modelcontextprotocol/sdk", "entry": "node node_modules/@cyanheads/crossref-mcp-server/dist/index.js", "env": { "CROSSREF_MAILTO": "(optional) [email protected] for polite-pool priority" }, "tools": [ { "name": "crossref_get_work", "params": { "doi": "string (required)" } }, { "name": "crossref_get_references", "params": { "doi": "string (required)" } }, { "name": "crossref_search_works", "params": { "query": "string", "filter": "object (e.g. {from-pub-date: '2025-01-01'})", "fields": "string[] (e.g. ['DOI','title','author'])", "rows": "number", "offset": "number", "cursor": "string", "sort": "string (e.g. 'is-referenced-by-count')", "order": "string ('asc'|'desc')" } }, { "name": "crossref_search_journals", "params": { "query": "string", "issn": "string", "include_works": "boolean", "rows": "number" } }, { "name": "crossref_search_funders", "params": { "query": "string", "funder_doi": "string", "include_works": "boolean", "rows": "number" } } ], "traces": [ { "call": "crossref_get_work", "args": { "doi": "10.1126/science.1225829" }, "ok": true, "ms": 186, "output_bytes": 2694, "note": "CRISPR paper — full metadata with 49K citations" }, { "call": "crossref_get_work", "args": { "doi": "10.1016/S0169-7552(98)00110-X" }, "ok": true, "ms": 181, "output_bytes": 792, "note": "PageRank paper — 8984 citations" }, { "call": "crossref_get_work", "args": { "doi": "10.48550/arXiv.1706.03762" }, "ok": false, "ms": 645, "note": "arXiv DOIs use DataCite, not Crossref — expected failure with recovery hint" }, { "call": "crossref_get_work", "args": { "doi": "10.1234/nonexistent-fake-doi-999" }, "ok": false, "ms": 182, "note": "invalid DOI — proper McpError with recovery hint" }, { "call": "crossref_get_references", "args": { "doi": "10.1126/science.1225829" }, "ok": true, "ms": 182, "output_bytes": 2954, "note": "47 references returned with DOIs and keys" }, { "call": "crossref_search_works", "args": { "query": "transformer attention mechanism", "rows": 3 }, "ok": true, "ms": 3733, "output_bytes": 862 }, { "call": "crossref_search_works", "args": { "query": "large language model", "filter": { "from-pub-date": "2025-01-01" }, "rows": 3 }, "ok": true, "ms": 3182, "output_bytes": 1022, "note": "date filter works" }, { "call": "crossref_search_works", "args": { "query": "deep learning", "sort": "is-referenced-by-count", "order": "desc", "rows": 3 }, "ok": true, "ms": 5016, "output_bytes": 913, "note": "ResNet paper returned first — correct sort by citations" }, { "call": "crossref_search_works", "args": { "query": "CRISPR gene editing", "fields": ["DOI", "title", "author", "is-referenced-by-count"], "rows": 2 }, "ok": true, "ms": 3594, "output_bytes": 202, "note": "field selection works — compact output" }, { "call": "crossref_search_journals", "args": { "query": "Nature" }, "ok": true, "ms": 648, "output_bytes": 1186 }, { "call": "crossref_search_journals", "args": { "issn": "0028-0836" }, "ok": true, "ms": 2002, "output_bytes": 138, "note": "ISSN lookup returns exact Nature match with 446K DOIs" }, { "call": "crossref_search_journals", "args": { "query": "Science", "include_works": true, "rows": 2 }, "ok": false, "ms": 1846, "note": "ambiguous journal — error with recovery hint to use ISSN instead" }, { "call": "crossref_search_funders", "args": { "query": "National Science Foundation" }, "ok": true, "ms": 2824, "output_bytes": 5166 }, { "call": "crossref_search_funders", "args": { "funder_doi": "10.13039/100000001" }, "ok": true, "ms": 5713, "output_bytes": 317, "note": "NSF — 449K+ works in Crossref" }, { "call": "crossref_search_funders", "args": { "query": "European Research Council", "include_works": true, "rows": 2 }, "ok": true, "ms": 9293, "output_bytes": 809 } ], "total_calls": 16, "success_rate": 0.75, "expected_failures": 4, "effective_success_rate": 1, "p50_ms": 2002, "p95_ms": 9293, "tested_at": "2026-06-26T10:17:29Z" }