@cyanheads/biorxiv-mcp-server 0.4.3 → 0.5.1

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (35) hide show
  1. package/AGENTS.md +20 -7
  2. package/CLAUDE.md +20 -7
  3. package/Dockerfile +14 -7
  4. package/README.md +24 -94
  5. package/changelog/0.5.x/0.5.0.md +23 -0
  6. package/changelog/0.5.x/0.5.1.md +22 -0
  7. package/dist/index.d.ts +1 -1
  8. package/dist/index.js +2 -1
  9. package/dist/index.js.map +1 -1
  10. package/dist/mcp-server/tools/definitions/biorxiv-list-recent.tool.d.ts +2 -0
  11. package/dist/mcp-server/tools/definitions/biorxiv-list-recent.tool.d.ts.map +1 -1
  12. package/dist/mcp-server/tools/definitions/biorxiv-list-recent.tool.js +18 -2
  13. package/dist/mcp-server/tools/definitions/biorxiv-list-recent.tool.js.map +1 -1
  14. package/dist/mcp-server/tools/definitions/biorxiv-search-preprints.tool.d.ts.map +1 -1
  15. package/dist/mcp-server/tools/definitions/biorxiv-search-preprints.tool.js +1 -1
  16. package/dist/mcp-server/tools/definitions/biorxiv-search-preprints.tool.js.map +1 -1
  17. package/dist/mcp-server/tools/definitions/index.d.ts +2 -0
  18. package/dist/mcp-server/tools/definitions/index.d.ts.map +1 -1
  19. package/dist/services/biorxiv/biorxiv-service.d.ts +3 -1
  20. package/dist/services/biorxiv/biorxiv-service.d.ts.map +1 -1
  21. package/dist/services/biorxiv/biorxiv-service.js +7 -8
  22. package/dist/services/biorxiv/biorxiv-service.js.map +1 -1
  23. package/dist/services/biorxiv/types.d.ts +8 -0
  24. package/dist/services/biorxiv/types.d.ts.map +1 -1
  25. package/dist/services/europe-pmc/europe-pmc-service.d.ts +5 -1
  26. package/dist/services/europe-pmc/europe-pmc-service.d.ts.map +1 -1
  27. package/dist/services/europe-pmc/europe-pmc-service.js +8 -3
  28. package/dist/services/europe-pmc/europe-pmc-service.js.map +1 -1
  29. package/dist/services/shared.d.ts +6 -2
  30. package/dist/services/shared.d.ts.map +1 -1
  31. package/dist/services/shared.js +38 -10
  32. package/dist/services/shared.js.map +1 -1
  33. package/package.json +4 -3
  34. package/server.json +3 -3
  35. package/skills/biorxiv-guide/SKILL.md +66 -0
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@cyanheads/biorxiv-mcp-server",
3
- "version": "0.4.3",
3
+ "version": "0.5.1",
4
4
  "mcpName": "io.github.cyanheads/biorxiv-mcp-server",
5
5
  "description": "Search and retrieve bioRxiv and medRxiv preprints — by DOI, date interval, or keyword — via MCP. STDIO or Streamable HTTP.",
6
6
  "type": "module",
@@ -18,7 +18,8 @@
18
18
  "CLAUDE.md",
19
19
  "AGENTS.md",
20
20
  "Dockerfile",
21
- "server.json"
21
+ "server.json",
22
+ "skills/"
22
23
  ],
23
24
  "scripts": {
24
25
  "build": "bun run scripts/build.ts",
@@ -95,7 +96,7 @@
95
96
  "postcss": "^8.5.23"
96
97
  },
97
98
  "dependencies": {
98
- "@cyanheads/mcp-ts-core": "^0.13.16",
99
+ "@cyanheads/mcp-ts-core": "^0.13.18",
99
100
  "defuddle": "^0.19.6",
100
101
  "linkedom": "^0.18.14",
101
102
  "pino-pretty": "^13.2.0",
package/server.json CHANGED
@@ -6,7 +6,7 @@
6
6
  "url": "https://github.com/cyanheads/biorxiv-mcp-server",
7
7
  "source": "github"
8
8
  },
9
- "version": "0.4.3",
9
+ "version": "0.5.1",
10
10
  "remotes": [
11
11
  {
12
12
  "type": "streamable-http",
@@ -19,7 +19,7 @@
19
19
  "registryBaseUrl": "https://registry.npmjs.org",
20
20
  "identifier": "@cyanheads/biorxiv-mcp-server",
21
21
  "runtimeHint": "npx",
22
- "version": "0.4.3",
22
+ "version": "0.5.1",
23
23
  "environmentVariables": [
24
24
  {
25
25
  "name": "BIORXIV_MAILTO",
@@ -58,7 +58,7 @@
58
58
  "registryBaseUrl": "https://registry.npmjs.org",
59
59
  "identifier": "@cyanheads/biorxiv-mcp-server",
60
60
  "runtimeHint": "npx",
61
- "version": "0.4.3",
61
+ "version": "0.5.1",
62
62
  "environmentVariables": [
63
63
  {
64
64
  "name": "MCP_TRANSPORT_TYPE",
@@ -0,0 +1,66 @@
1
+ ---
2
+ name: biorxiv-guide
3
+ description: Search syntax, identifier origins, and counting gates for the biorxiv_* tools over bioRxiv and medRxiv preprints. Use when finding preprints by topic, author, or date window; counting what a category, funder, or journal took up; tracing a preprint to its journal article or back; or reading a preprint's full text or an earlier revision.
4
+ license: Apache-2.0
5
+ metadata:
6
+ author: cyanheads
7
+ version: "1.1"
8
+ ---
9
+
10
+ ## Scope
11
+
12
+ bioRxiv (biology) and medRxiv (health sciences) preprints: metadata, every revision, the preprint-to-journal crosswalk, and full text. Keyless. Listings and DOI lookups read api.biorxiv.org; keyword search is EuropePMC's ranked index, restricted to the two servers, so arXiv, Research Square, SSRN, and other preprint servers never appear.
13
+
14
+ No citation counts and no journal-article content beyond the crosswalk record. Pair with a PubMed, Crossref, or OpenAlex server for the journal side, and with an arXiv server for computational biology posted only there.
15
+
16
+ ## Identifiers
17
+
18
+ - **DOI prefix is an era, not a server.** `10.1101/` is the older prefix and `10.64898/` the newer; bioRxiv and medRxiv both issue both, and a preprint keeps its DOI through every revision (`10.1101/2025.11.24.690249` was revised as v2 under `10.1101/` long after new submissions moved to `10.64898/`). Leave `server` at `"both"` for DOI lookups; the response names the server that holds it.
19
+ - **The date inside a DOI is the submission date, not the posting date.** `10.1101/2024.01.03.573434` was posted 2024-01-04; `10.1101/2025.05.22.653117` on 2025-05-27. Read `date` (or `preprintDate` on crosswalk rows) for when a preprint appeared. Older bioRxiv DOIs are plain serials with no date (`10.1101/049882`).
20
+ - **Versions.** Each revision has its own title, authors, and date; `biorxiv_get_preprint` lists them. Only the latest revision carries an abstract — the API keeps one per preprint. An earlier revision's own abstract and text come from `biorxiv_get_fulltext` with `version`: `10.1101/049882` v1 opens with a different abstract than the one `biorxiv_get_preprint` returns.
21
+ - **Category strings** come from `biorxiv_list_categories`, or from any record's `category` field, which is accepted as input as written.
22
+ - **Funder** is a ROR ID (`https://ror.org/<id>`), looked up by name at ror.org. A Crossref Funder Registry ID (`100000002`) is not one and is rejected before any request.
23
+ - **Publisher** is the DOI prefix of a journal publisher, the part before the slash, taken from any article DOI of the journal you care about. It names the whole publisher, not one journal: `10.1038` returns rows from many Springer Nature journals at once (`s41467-`, `s41598-`, `s41593-`, and more DOI stems on one two-day page).
24
+
25
+ ## Gates
26
+
27
+ ### Keyword search (`biorxiv_search_preprints`)
28
+
29
+ `query` is EuropePMC query syntax, passed through as written inside one pair of parentheses, then joined to the server's author, date, and server clauses with `AND`.
30
+
31
+ - **Spell out synonyms yourself.** Synonym expansion is off, so `"heart attack"` misses the far larger set written `"myocardial infarction"`. Search `"heart attack" OR "myocardial infarction"`.
32
+ - **Scope by field when precision matters.** A bare term or phrase matches more than the title and abstract. `TITLE_ABS:"…"`, `TITLE:"…"`, `ABSTRACT:"…"`, `AUTH:"…"`, and `AFF:"…"` all work inside `query`; `TITLE_ABS:"base editing"` and `TITLE:"base editing"` each match fewer records than the bare phrase.
33
+ - **Author names.** `author` becomes `AUTH:"…"`. A full name and `Surname Initial` match the same records (`Jennifer Doudna`, `Doudna J`); adding a middle initial (`Doudna JA`) drops records indexed without it. The surname alone is broadest. Confirm identity from co-authors and institution before attributing a list to one person.
34
+ - **`server` scopes the search itself**, not only enrichment: `"medrxiv"` returns medRxiv preprints only.
35
+ - **The date window is first posting; the record shows the latest revision.** `date_from` / `date_to` filter on when v1 appeared, while each result's `date` and `version` are the latest revision's: a January 2024 window returned `10.1101/2024.01.29.577624` dated 2024-12-15, v2. Never re-filter search results on `date` to enforce the window.
36
+
37
+ ### Posting-date listing (`biorxiv_list_recent`)
38
+
39
+ - **Rows are revisions, and `total` counts them.** A revised preprint reappears on the day its new version posts, under its old DOI and `version` 2, 3, …. bioRxiv's 2026-10-01..07 listing held 1,460 rows, of which 1,092 were new preprints. To count new preprints, read each server's `newPreprints` from the first page, which carries the whole listing's count for those filters; to list them, keep `version` `"1"` rows.
40
+ - **It has no keyword filter.** At 30 rows a page, one bioRxiv week runs to dozens of pages. For a topic within a window, search with `date_from` / `date_to` and filter enriched results on `category` locally; reserve the listing for a complete category census or for the last day or two that EuropePMC has not indexed yet.
41
+ - **A funder filter is not a funder's output.** It reaches only records api.biorxiv.org linked to that ROR ID: NIH (`01cwqze88`) returned 9 rows for that same week, all of them revisions. Report what it returns as "preprints bioRxiv links to this funder", never as everything the funder supported.
42
+
43
+ ### Journal crosswalk
44
+
45
+ - **A missing `publishedJournalDoi` means bioRxiv records no journal version**, which is not proof that none exists. Before reporting a preprint as unpublished, search the journal literature by its authors and key title words on a PubMed or Crossref server; titles drift, as `10.1101/049882`'s did across its five revisions.
46
+ - **medRxiv's `type` reads `PUBLISHAHEADOFPRINT` on its preprints**, published or not. It says nothing about journal publication; `publishedJournalDoi` does.
47
+ - **Publication listings count rows, not preprints.** medRxiv can list one preprint twice with different `preprintDate` values (one week's first page: 100 rows, 97 distinct preprints). Deduplicate on `preprintDoi` before counting.
48
+ - **A past interval's total keeps growing** as bioRxiv records more matches (the medRxiv week 2026-09-01..07 held 103 rows on 2026-10-17 and 107 on 2026-10-28). State counts as of the day you ran them, and finish a walk in one session.
49
+
50
+ ## Workflows
51
+
52
+ **New preprints on a topic in a period.** `biorxiv_search_preprints` with a field-scoped query and `date_from` / `date_to`, `include_abstract: false` and `limit: 100` for triage, passing each `nextCursorMark` back as `cursor_mark`. Narrow to a field by the `category` on enriched results (EuropePMC-fallback results carry none). Read abstracts for the shortlist with `biorxiv_get_preprint` (10 DOIs a call). For the last day or two, add a `biorxiv_list_recent` pass with `category`.
53
+
54
+ **Everything a category posted in a window.** `biorxiv_list_recent` with `category`, each server advanced at its own `nextCursor` until none is returned; keep `version` `"1"` rows for new preprints, whose number each server's `newPreprints` gives on the first page.
55
+
56
+ **Find a person's preprints.** `author` (surname plus initial, then surname alone if the list looks short) with `server` when the field is clear; add a topic `query` to separate namesakes.
57
+
58
+ **Has this preprint been published, and where?** `biorxiv_get_preprint` shows `publishedJournalDoi`; `biorxiv_get_published_version` adds the journal name and date. Holding the journal article instead, pass its DOI to `biorxiv_get_published_version` to find the preprint. A not-found answer is bioRxiv's record only, per the crosswalk gate.
59
+
60
+ **What a journal or publisher took up in a period.** `biorxiv_list_published_versions` filters on the journal's publication date, so the preprints can be years older.
61
+
62
+ - One bioRxiv publisher: pass `publisher` and walk with `limit: 100`. Rows carry no journal name, so isolate one journal by its DOI stem in `publishedDoi` (`10.1038/s41422-` is Cell Research), then call `biorxiv_get_published_version` for the full record of rows you keep.
63
+ - medRxiv, or a journal named in words: list without `publisher` on the one server and match `publishedJournal`, which those rows carry.
64
+ - Deduplicate on `preprintDoi` before counting.
65
+
66
+ **Read a preprint.** `biorxiv_get_fulltext` returns sections as `## ` headings; every chunk reports `totalChars`, and a `limit` at or above it from `offset` 0 returns the whole text in one call. Table cells are never in the text; `omittedTables` gives the page that serves each table's values. To compare what the preprint claimed at first with what it claims now, read `version: 1` beside the latest.