@cyanheads/biorxiv-mcp-server 0.3.0 → 0.4.0

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 (55) hide show
  1. package/AGENTS.md +40 -31
  2. package/CLAUDE.md +40 -31
  3. package/Dockerfile +85 -40
  4. package/README.md +34 -44
  5. package/changelog/0.3.x/0.3.1.md +33 -0
  6. package/changelog/0.4.x/0.4.0.md +33 -0
  7. package/dist/index.js +1 -1
  8. package/dist/index.js.map +1 -1
  9. package/dist/mcp-server/tools/definitions/biorxiv-get-fulltext.tool.d.ts +7 -2
  10. package/dist/mcp-server/tools/definitions/biorxiv-get-fulltext.tool.d.ts.map +1 -1
  11. package/dist/mcp-server/tools/definitions/biorxiv-get-fulltext.tool.js +29 -15
  12. package/dist/mcp-server/tools/definitions/biorxiv-get-fulltext.tool.js.map +1 -1
  13. package/dist/mcp-server/tools/definitions/biorxiv-get-preprint.tool.d.ts +1 -1
  14. package/dist/mcp-server/tools/definitions/biorxiv-get-preprint.tool.d.ts.map +1 -1
  15. package/dist/mcp-server/tools/definitions/biorxiv-get-preprint.tool.js +4 -6
  16. package/dist/mcp-server/tools/definitions/biorxiv-get-preprint.tool.js.map +1 -1
  17. package/dist/mcp-server/tools/definitions/biorxiv-get-published-version.tool.d.ts +1 -1
  18. package/dist/mcp-server/tools/definitions/biorxiv-get-published-version.tool.d.ts.map +1 -1
  19. package/dist/mcp-server/tools/definitions/biorxiv-get-published-version.tool.js +4 -6
  20. package/dist/mcp-server/tools/definitions/biorxiv-get-published-version.tool.js.map +1 -1
  21. package/dist/mcp-server/tools/definitions/biorxiv-list-published-versions.tool.d.ts +95 -0
  22. package/dist/mcp-server/tools/definitions/biorxiv-list-published-versions.tool.d.ts.map +1 -0
  23. package/dist/mcp-server/tools/definitions/biorxiv-list-published-versions.tool.js +322 -0
  24. package/dist/mcp-server/tools/definitions/biorxiv-list-published-versions.tool.js.map +1 -0
  25. package/dist/mcp-server/tools/definitions/biorxiv-list-recent.tool.d.ts +1 -1
  26. package/dist/mcp-server/tools/definitions/biorxiv-list-recent.tool.d.ts.map +1 -1
  27. package/dist/mcp-server/tools/definitions/biorxiv-list-recent.tool.js +7 -11
  28. package/dist/mcp-server/tools/definitions/biorxiv-list-recent.tool.js.map +1 -1
  29. package/dist/mcp-server/tools/definitions/biorxiv-search-preprints.tool.d.ts +4 -4
  30. package/dist/mcp-server/tools/definitions/biorxiv-search-preprints.tool.d.ts.map +1 -1
  31. package/dist/mcp-server/tools/definitions/biorxiv-search-preprints.tool.js +18 -27
  32. package/dist/mcp-server/tools/definitions/biorxiv-search-preprints.tool.js.map +1 -1
  33. package/dist/mcp-server/tools/definitions/index.d.ts +81 -8
  34. package/dist/mcp-server/tools/definitions/index.d.ts.map +1 -1
  35. package/dist/mcp-server/tools/definitions/index.js +3 -0
  36. package/dist/mcp-server/tools/definitions/index.js.map +1 -1
  37. package/dist/services/biorxiv/biorxiv-service.d.ts +17 -2
  38. package/dist/services/biorxiv/biorxiv-service.d.ts.map +1 -1
  39. package/dist/services/biorxiv/biorxiv-service.js +61 -1
  40. package/dist/services/biorxiv/biorxiv-service.js.map +1 -1
  41. package/dist/services/biorxiv/types.d.ts +24 -3
  42. package/dist/services/biorxiv/types.d.ts.map +1 -1
  43. package/dist/services/biorxiv-fulltext/biorxiv-fulltext-service.d.ts +17 -2
  44. package/dist/services/biorxiv-fulltext/biorxiv-fulltext-service.d.ts.map +1 -1
  45. package/dist/services/biorxiv-fulltext/biorxiv-fulltext-service.js +86 -7
  46. package/dist/services/biorxiv-fulltext/biorxiv-fulltext-service.js.map +1 -1
  47. package/dist/services/europe-pmc/europe-pmc-service.d.ts.map +1 -1
  48. package/dist/services/europe-pmc/europe-pmc-service.js +3 -2
  49. package/dist/services/europe-pmc/europe-pmc-service.js.map +1 -1
  50. package/dist/services/shared.d.ts +5 -3
  51. package/dist/services/shared.d.ts.map +1 -1
  52. package/dist/services/shared.js +19 -10
  53. package/dist/services/shared.js.map +1 -1
  54. package/package.json +13 -12
  55. package/server.json +10 -24
package/AGENTS.md CHANGED
@@ -1,11 +1,11 @@
1
1
  # Agent Protocol
2
2
 
3
3
  **Server:** biorxiv-mcp-server
4
- **Version:** 0.3.0
5
- **Framework:** [@cyanheads/mcp-ts-core](https://www.npmjs.com/package/@cyanheads/mcp-ts-core) `^0.13.6`
4
+ **Version:** 0.4.0
5
+ **Framework:** [@cyanheads/mcp-ts-core](https://www.npmjs.com/package/@cyanheads/mcp-ts-core) `^0.13.15`
6
6
  **Engines:** Bun ≥1.4.0, Node ≥24.0.0
7
- **MCP SDK:** `@modelcontextprotocol/server` ^2.0.0
8
- **Zod:** ^4.6.5
7
+ **MCP SDK:** `@modelcontextprotocol/server` ^2.3.1
8
+ **Zod:** ^4.6.6
9
9
 
10
10
  > **Read the framework docs first:** `node_modules/@cyanheads/mcp-ts-core/CLAUDE.md` contains the full API reference — builders, Context, error codes, exports, patterns. This file covers server-specific conventions only.
11
11
 
@@ -105,8 +105,18 @@ export function getServerConfig(): z.infer<typeof ServerConfigSchema> {
105
105
 
106
106
  `parseEnvConfig` maps Zod schema paths → env var names so validation errors name the actual variable rather than the internal path. It reads an empty string and an unsubstituted whole-value `${…}` placeholder (an MCPB optional field left blank) as unset, so an optional field stays `undefined` and a defaulted field takes its default — no per-field `z.preprocess` guard needed.
107
107
 
108
+ For env booleans use `z.stringbool()`, never `z.coerce.boolean()` — `Boolean("false")` is `true`, so a coerced flag can't be disabled through the environment. `z.stringbool()` parses `true/false/1/0/yes/no/on/off` and rejects anything else, so `=false` actually disables.
109
+
110
+ ### Server identity and instructions
111
+
112
+ `createApp()` accepts optional identity fields forwarded to the SDK's `initialize` response and the server manifest (`/.well-known/mcp.json`): `name`, `title`, `websiteUrl`, `description` (wins over `MCP_SERVER_DESCRIPTION`), `icons`, and `instructions`. This server sets `name` and `title` (both the repo name) and `instructions` in `src/index.ts`.
113
+
114
+ `instructions` is optional server-level orientation, sent on every `initialize` as session-level context. Use it for deployment guidance (connection aliases, regional notes, scope hints) instead of repeating the same context across tool descriptions. Client adoption is uneven, but there's no downside when set.
115
+
108
116
  ### Session posture and shutdown
109
117
 
118
+ Two more `createApp()` options shape how the server runs rather than how it presents itself:
119
+
110
120
  ```ts
111
121
  await createApp({
112
122
  sessionMode: 'stateless', // or { default: 'stateful', require: 'stateful' }
@@ -115,9 +125,9 @@ await createApp({
115
125
  });
116
126
  ```
117
127
 
118
- `sessionMode` declares the HTTP session posture in `src/` instead of leaving it to a deployment's `MCP_SESSION_MODE`, which still wins whenever it carries a meaningful value. This server declares `stateless` — no tool calls `ctx.requestInput`. Add `require: 'stateful'` if a tool ever asks the caller for input mid-handler: startup then fails with a `ConfigurationError` rather than serving a mode in which a 2025-era client can never answer the prompt. Stdio is never refused.
128
+ `sessionMode` declares the HTTP session posture in `src/` instead of leaving it to a deployment's `MCP_SESSION_MODE`, which still wins whenever it carries a meaningful value (an empty string and an unsubstituted `${…}` placeholder read as unset and fall through to the option). This server declares `stateless` — no tool calls `ctx.requestInput`. Add `require: 'stateful'` if a tool ever asks the caller for input mid-handler: startup then fails with a `ConfigurationError` rather than serving a mode in which a 2025-era client can never answer the prompt. Stdio is never refused.
119
129
 
120
- `teardown(core)` is the `setup()` counterpart — release a watcher, socket, or non-`unref()`'d timer there. It runs after the transport stops and before the logger closes, on every shutdown path. This server's services allocate none, so it declares no `teardown`.
130
+ `teardown(core)` is the `setup()` counterpart — release a watcher, socket, or non-`unref()`'d timer there. It runs after the transport stops and before the logger closes, on every shutdown path, and a signal-triggered shutdown then exits the process explicitly (0, or 1 if a step never settles within the framework's 10 s ceiling). This server's services allocate none, so it declares no `teardown`.
121
131
 
122
132
  ---
123
133
 
@@ -128,14 +138,14 @@ Handlers receive a unified `ctx` object. Key properties:
128
138
  | Property | Description |
129
139
  |:---------|:------------|
130
140
  | `ctx.log` | Request-scoped logger — `.debug()`, `.info()`, `.notice()`, `.warning()`, `.error()`. Auto-correlates requestId, traceId, tenantId. Dual-sink: Pino **and** `notifications/message` to the client, so treat it as client-visible. |
131
- | `ctx.state` | Tenant-scoped KV — `.get(key)`, `.set(key, value, { ttl? })`, `.delete(key)`, `.getMany(keys)`, `.list(prefix, { cursor, limit })`. Accepts any serializable value. |
141
+ | `ctx.state` | Tenant-scoped KV — `.get(key)`, `.set(key, value, { ttl? })`, `.delete(key)`, `.getMany(keys)`, `.list(prefix, { cursor, limit })`. Accepts any JSON-serializable value; reads return its JSON form (a `Date` comes back as an ISO string). |
132
142
  | `ctx.requestInput` | Suspend and ask the caller for more input — `return ctx.requestInput({ inputRequests: { key: inputRequired.elicit({ message, requestedSchema }) } })`. Never returns; the handler is re-entered with the answers. Always present. |
133
- | `ctx.inputs` | Reader over a retried request's responses — `.accepted(key, schema)`, `.view(key)`, `.state()`, `.dropped`. Empty on the first round. |
134
- | `ctx.recoveryFor(reason)` | Typed lookup of the contract `recovery` for a declared reason. Returns `{ recovery: { hint } }` for known reasons, `{}` otherwise. Spread into `ctx.fail` data to mirror the contract hint into `content[]`. |
143
+ | `ctx.inputs` | The request's responses — `.accepted(key, schema)`, `.view(key)`, `.state()`, `.dropped` — limited to what the client declared (`elicitation` and its form/url modes, `sampling`, `roots`). Client-supplied: a consent gate trusts only a `ctx.state` record it stored when it asked, bound to the operation, caller, and target (see the `api-context` skill). |
144
+ | `ctx.clientCapabilities` | What the client declared for this request, `undefined` when no view exists. Decides whether to ask for optional context (e.g. roots); never a reason to skip a consent prompt. |
135
145
  | `ctx.enrich` | Success-path agent context (empty-result notices, query echo, pagination totals) — `ctx.enrich(...)` or `.notice()` / `.total()` / `.echo()` / `.truncated()`. Reaches `structuredContent` and `content[]`; lands only when the definition declares an `enrichment` block (no-op otherwise). |
136
146
  | `ctx.content` | Non-text content blocks — `.image(data, mimeType)`, `.audio(data, mimeType)`, or `ctx.content(block)` for a raw block. Prepended to `content[]` after `format()`; never enters `structuredContent`. |
137
147
  | `ctx.signal` | `AbortSignal` for cancellation. |
138
- | `ctx.requestId` | Unique request ID. |
148
+ | `ctx.requestId` | Request ID — the one every log record of the call carries and its error envelope returns as `data.requestId`. |
139
149
  | `ctx.tenantId` | Tenant ID from JWT; `'default'` for stdio or HTTP with auth off. |
140
150
 
141
151
  ---
@@ -144,7 +154,7 @@ Handlers receive a unified `ctx` object. Key properties:
144
154
 
145
155
  Handlers throw — the framework catches, classifies, and formats.
146
156
 
147
- **Recommended: typed error contract.** Declare `errors: [{ reason, code, when, recovery, retryable?, severity?, thrownBy? }]` on `tool()` / `resource()` to receive a typed `ctx.fail(reason, …)` keyed by the declared reason union. TypeScript catches `ctx.fail('typo')` at compile time, `data.reason` is auto-populated for observability, and the linter enforces conformance against the handler body. The `recovery` field is required (≥ 5 words, lint-validated). Spread `ctx.recoveryFor('reason')` into `data` to mirror the contract recovery onto the wire (`data.recovery.hint`, mirrored into `content[]` text unless the message already contains it verbatim), or pass an explicit `recovery: { hint }` when the hint needs runtime context (a computed wait). Forwarding is lint-enforced per throw site — a `ctx.fail` carrying neither warns as `error-contract-recovery-unforwarded`. Every declared reason must be named by a literal `ctx.fail('<reason>'` or `ctx.recoveryFor('<reason>')` in its handler — `lint:mcp`'s `error-contract-unthrown` flags one that isn't; mark an entry the service layer throws with `thrownBy: 'service'` (lint-only metadata) instead. Baseline codes (`InternalError`, `ServiceUnavailable`, `Timeout`, `ValidationError`, `SerializationError`, `RequestCancelled`) bubble freely and don't need declaring.
157
+ **Recommended: typed error contract.** Declare `errors: [{ reason, code, when, recovery, retryable?, severity?, thrownBy? }]` on `tool()` / `resource()` to receive `ctx.fail(reason, …)` typed against the reason union. TypeScript catches typos at compile time, `data.reason` is auto-populated for observability, linter enforces conformance against the handler body. `recovery` is required (≥ 5 words, lint-validated) — the single source of truth for the agent's next move. The framework puts it on the wire whenever a failure carrying that `reason` arrives without a hint — a bare `ctx.fail('reason')` or a service throw with `data: { reason }` — as `data.recovery.hint`, mirrored into `content[]` text unless the message already contains it verbatim; override with an explicit `{ recovery: { hint: '...' } }` when dynamic runtime context matters (a computed wait). Every error envelope also carries `data.requestId`, the id the server's log records for that call carry, and `content[]` closes with `(reason … · request <id>)`. Every declared reason must be named by a literal `ctx.fail('<reason>'` or `ctx.recoveryFor('<reason>')` in its handler — `lint:mcp`'s `error-contract-unthrown` flags one that isn't; mark an entry the service layer throws with `thrownBy: 'service'` so the rule skips it — lint-only metadata, nothing at runtime reads it. Baseline codes (`InternalError`, `ServiceUnavailable`, `Timeout`, `ValidationError`, `SerializationError`, `RequestCancelled`) bubble freely and don't need declaring.
148
158
 
149
159
  ```ts
150
160
  import { JsonRpcErrorCode } from '@cyanheads/mcp-ts-core/errors';
@@ -157,9 +167,7 @@ errors: [
157
167
  async handler(input, ctx) {
158
168
  const result = await getBiorxivApiService().getDetails(input.doi, input.server ?? 'both');
159
169
  if (result.collection.length === 0) {
160
- throw ctx.fail('doi_not_found', `DOI not found: ${input.doi}`, {
161
- ...ctx.recoveryFor('doi_not_found'),
162
- });
170
+ throw ctx.fail('doi_not_found', `DOI not found: ${input.doi}`);
163
171
  }
164
172
  return result;
165
173
  }
@@ -167,7 +175,7 @@ async handler(input, ctx) {
167
175
 
168
176
  **Declare contracts inline on each tool.** The contract is part of the tool's public surface — one file should give the full picture. Don't extract a shared `errors[]` constant; per-tool repetition is the intended cost of locality.
169
177
 
170
- **Fallback (no contract entry fits):** error factories or plain `Error`.
178
+ **Fallback (no contract entry fits):** throw via factories or plain `Error`.
171
179
 
172
180
  ```ts
173
181
  import { notFound, serviceUnavailable } from '@cyanheads/mcp-ts-core/errors';
@@ -175,7 +183,7 @@ throw notFound('Preprint not found', { doi });
175
183
  throw serviceUnavailable('api.biorxiv.org unreachable', { url }, { cause: err });
176
184
  ```
177
185
 
178
- See framework CLAUDE.md and the `api-errors` skill for the full auto-classification table, all factories, and the contract reference.
186
+ See framework CLAUDE.md and the `api-errors` skill for the full auto-classification table, all available factories, and the contract reference.
179
187
 
180
188
  ---
181
189
 
@@ -205,6 +213,7 @@ src/
205
213
  biorxiv-list-recent.tool.ts
206
214
  biorxiv-search-preprints.tool.ts
207
215
  biorxiv-get-published-version.tool.ts
216
+ biorxiv-list-published-versions.tool.ts
208
217
  biorxiv-get-fulltext.tool.ts
209
218
  biorxiv-list-categories.tool.ts
210
219
  ```
@@ -228,12 +237,12 @@ src/
228
237
  - **Category spelling:** the listing API filters only on lowercase with `/` and `_` read as a space (`hiv aids`), and answers anything else with its unfiltered listing, echoing `messages[0].category: "all"`. `BiorxivApiService` owns the spelling and flags `categoryIgnored`; never present a flagged page as filtered.
229
238
  - **Funder data:** upstream funder `name`s and ROR `id`s on records are misattributed — expose only `awards` (verbatim, never split), never a funder name or ID. The listing's `funder` filter takes a bare ROR ID (`normalizeRorId()` strips the URL form and checks the checksum), is bioRxiv-only, and `"funder value not found"` (flagged `funderNotFound`) is an error, never an empty page.
230
239
  - **Server parameter:** `"biorxiv" | "medrxiv" | "both"`. Default to `"both"`, including for DOI-resolution tools — both servers issue `10.1101/` DOIs, so the DOI never identifies the server and a single-server default silently misses records held by the other. Fan out via `Promise.allSettled` when `"both"` and name the answering server in the output.
231
- - **Pagination:** integer `cursor` offset (0, 30, 60, …). Page size is fixed at 30 by the API — do not expose a `limit` parameter.
232
- - **Two-server pagination:** when `server="both"`, surface per-server state: `{ biorxiv: { cursor, total }, medrxiv: { cursor, total } }`. A merged cursor is ambiguous.
240
+ - **Pagination:** integer `cursor` offset. Page size is fixed by the API — 30 on `biorxiv_list_recent` (0, 30, 60, …), 100 on `biorxiv_list_published_versions` (0, 100, 200, …) — do not expose a `limit` parameter.
241
+ - **Two-server pagination:** when `server="both"`, surface per-server state: `{ biorxiv: { cursor, total }, medrxiv: { cursor, total } }`. A merged cursor is ambiguous. A server whose cursor overshot is flagged `exhausted`: the posting-date listing reports `total: 0` there, and the publication-date listing reports no total at all, so its exhausted entry carries none.
233
242
  - **EuropePMC enrichment:** `biorxiv_search_preprints` uses EuropePMC for relevance then enriches matching DOIs via the details endpoint. Enrichment routes based on `input.server` — when explicit ("biorxiv" or "medrxiv"), always use that server. `10.1101/` prefix is used as a hint only within `server="both"` to prefer bioRxiv first, but is not a reliable discriminator (both servers share this prefix). The ranked search stays `resulttype=lite`, which never carries an abstract; records that fall back to EuropePMC metadata get theirs from one DOI-keyed `resultType=core` query (`getAbstracts`), made only when abstracts are requested and something fell back. A failed lookup is a notice, never a failed search.
234
- - **`format()` must be content-complete:** Claude Desktop reads `content[]` from `format()`, not `structuredContent`. Revision list, crosswalk data, pagination state, and every abstract a record carries must all appear in the rendered markdown, not just counts. `include_abstract` (default `false` on `biorxiv_list_recent`, `true` on `biorxiv_search_preprints`) removes the abstract from both surfaces together and no other field.
243
+ - **`format()` must be content-complete:** Claude Desktop reads `content[]` from `format()`, not `structuredContent`. Revision list, crosswalk data, pagination state, and every abstract a record carries must all appear in the rendered markdown, not just counts. `include_abstract` (default `false` on `biorxiv_list_recent` and `biorxiv_list_published_versions`, `true` on `biorxiv_search_preprints`) removes the abstract from both surfaces together and no other field.
235
244
  - **Upstream text in `format()`:** run titles and abstracts through `normalizeUpstreamText()` in the service (never in `format()`), and pass every upstream free-text field through `escapeMarkdown()` where `format()` interpolates it. It escapes only characters that would change rendering (`A\*T`, `\<COMP id>`; `p<0.05` and `cc_by` stay raw), because agents copy spans out of `content[]` — keep it that narrow. Never escape DOIs, URLs, `structuredContent`, or `biorxiv_get_fulltext`'s extracted `content`.
236
- - **Polite access:** include `BIORXIV_MAILTO` in the `User-Agent` header when set: `biorxiv-mcp-server/0.1.0 (mailto:${config.mailto})`. The env var is optional — omit the mailto segment when not configured.
245
+ - **Polite access:** include `BIORXIV_MAILTO` in the `User-Agent` header when set: `biorxiv-mcp-server/${SERVER_VERSION} (mailto:${config.mailto})`. The env var is optional — omit the mailto segment when not configured.
237
246
 
238
247
  ---
239
248
 
@@ -265,21 +274,21 @@ Available skills:
265
274
  | `release-and-publish` | Fast-forward merge (release PR mode) + tag + push + npm + MCP Registry + GH Release + Docker. Picks up from `git-wrapup` |
266
275
  | `maintenance` | Investigate changelogs, adopt upstream changes, sync skills to agent dirs |
267
276
  | `orchestrations` | Chain task skills into a gated multi-phase pipeline — build-out, QA-fix, update-ship — when you can spawn sub-agents |
277
+ | `report-issue-framework` | File a bug or feature request against `@cyanheads/mcp-ts-core` via `gh` CLI |
278
+ | `report-issue-local` | File a bug or feature request against this server's own repo via `gh` CLI |
279
+ | `techniques` | Catalog of response/data-shaping techniques — overflow handling, payload shaping, retrieval patterns |
268
280
  | `api-auth` | Auth modes, scopes, JWT/OAuth |
269
281
  | `api-canvas` | DataCanvas: register tabular data, run SQL, export, plus the `spillover()` helper for big result sets — Tier 3 opt-in |
270
282
  | `api-config` | AppConfig, parseConfig, env vars |
271
283
  | `api-context` | Context interface, RequestContext, logger, state, multi-round-trip input |
272
284
  | `api-errors` | McpError, JsonRpcErrorCode, error patterns |
273
285
  | `api-linter` | Definition linter rule catalog — invoked by `bun run lint:mcp` and `devcheck` |
274
- | `api-mirror` | MirrorService: persistent, self-refreshing local mirror of a bulk upstream dataset (embedded SQLite + FTS5) — Tier 3 opt-in |
286
+ | `api-mirror` | MirrorService: persistent self-refreshing local mirror (embedded SQLite + FTS5) of a bulk upstream dataset — Tier 3 opt-in |
275
287
  | `api-services` | LLM, Speech, Graph services |
276
- | `api-telemetry` | OTel catalog: spans, metrics, completion logs, env config, cardinality rules |
277
288
  | `api-testing` | createMockContext, test patterns |
278
289
  | `api-utils` | Formatting, parsing, security, pagination, scheduling, telemetry helpers |
290
+ | `api-telemetry` | OTel catalog: spans, metrics, completion logs, env config, cardinality rules |
279
291
  | `api-workers` | Cloudflare Workers runtime |
280
- | `report-issue-framework` | File a bug or feature request against `@cyanheads/mcp-ts-core` via `gh` CLI |
281
- | `report-issue-local` | File a bug or feature request against this server's own repo via `gh` CLI |
282
- | `techniques` | Catalog of response/data-shaping techniques — overflow handling, payload shaping, retrieval patterns |
283
292
 
284
293
  **Chaining skills into pipelines.** When the user wants a multi-phase effort — build this server out, QA-and-fix the surface, update-and-ship — *and you can spawn sub-agents*, `framework-skills/orchestrations/SKILL.md` sequences the task skills above into a gated pipeline with verification at each step. Read it to drive the run. Optional: skip it if you can't orchestrate sub-agents, and ignore it entirely if you were *spawned* as one — you've already been scoped to a single phase.
285
294
 
@@ -299,18 +308,18 @@ When you complete a skill's checklist, check the boxes and add a completion time
299
308
  | `bun run devcheck` | Lint + format + typecheck + security + changelog sync |
300
309
  | `bun run audit:fix` | `bun audit fix` — upgrade vulnerable packages to the lowest safe version within existing ranges (`--dry-run` previews, `--latest` rewrites ranges). First response when `devcheck` flags a transitive advisory; then `bun update <name>`, then `bun dedupe` |
301
310
  | `bun run audit:refresh` | Delete `bun.lock` and reinstall. Last resort after `audit:fix`, `bun update <name>`, and `bun dedupe` — re-resolves every ranged dep (the framework pin included) and rewrites the lockfile as `lockfileVersion: 2` |
311
+ | `bun run lint:mcp` | Run the MCP definition linter standalone (rule catalog: `api-linter` skill) |
312
+ | `bun run lint:packaging` | Packaging surface checks — `server.json`/`manifest.json` env-var parity (run by devcheck) |
313
+ | `bun run list-skills` | Print the skill registry |
302
314
  | `bun run tree` | Generate directory structure doc |
303
315
  | `bun run format` | Auto-fix formatting (safe fixes only) |
304
316
  | `bun run format:unsafe` | Also apply Biome's unsafe autofixes — review the diff; they can change behavior |
305
317
  | `bun run test` | Run tests (Vitest — use `bun run test`, not `bun test`) |
306
- | `bun run lint:mcp` | Run the MCP definition linter standalone (rule catalog: `api-linter` skill) |
307
- | `bun run lint:packaging` | Packaging surface checks — `server.json`/`manifest.json` env-var parity (run by devcheck) |
308
- | `bun run bundle` | Build, pack, and clean a `.mcpb` for one-click Claude Desktop install |
309
- | `bun run list-skills` | Surface available local skills for sub-agents |
310
318
  | `bun run start:stdio` | Production mode (stdio) |
311
319
  | `bun run start:http` | Production mode (HTTP) |
312
320
  | `bun run changelog:build` | Regenerate `CHANGELOG.md` from `changelog/*.md` |
313
321
  | `bun run changelog:check` | Verify `CHANGELOG.md` is in sync (used by devcheck) |
322
+ | `bun run bundle` | Build, pack, and clean a `.mcpb` for one-click Claude Desktop install |
314
323
  | `bun run release:github` | Create (or repair) a GitHub Release on the current version's annotated tag |
315
324
 
316
325
  **CI is one file.** `.github/workflows/codeql.yml` is the only GitHub Actions workflow: CodeQL is GitHub-owned end to end, and the file runs only while the repo's CodeQL *default setup* is turned off. Verification — `devcheck`, tests, the release gates — runs locally; don't add a workflow that re-runs it.
@@ -356,7 +365,7 @@ security: false # optional — true ONLY for a source
356
365
 
357
366
  ## Publishing
358
367
 
359
- **Every release goes through a gated release PR** — `git-wrapup`'s "Release PR mode", mode `gated`. Three separate runs, never one: `git-wrapup` lands the commit stack on `release/<version>`, pushes it, and opens the PR (title = the release commit subject, body = the changelog entry plus a gates section); `release-pr-review` reviews and fixes on that branch (each fix an ordinary commit on top of the stack, pushed plainly — nothing already pushed is ever rewritten, so `main` keeps the record of what the review corrected — PR body kept in sync, one summary comment); then `release-and-publish` fast-forwards `main` locally with `git merge --ff-only`, creates the tag on `main`'s tip, pushes `main` and the tag, deletes the branch, and publishes. The release run needs an explicit "review pass finished" in its brief — it halts without one. **Never merge through the GitHub UI or `gh pr merge`**: squash and rebase-merge are disabled in the repo settings because both rewrite the stack (rebase-merge also strips the SSH signatures), and a merge commit breaks the linear history.
368
+ **Every release goes through a gated release PR** — `git-wrapup`'s "Release PR mode", mode `gated`. Three separate runs, never one: `git-wrapup` lands the commit stack on `release/<version>`, pushes it, and opens the PR (title = the release commit subject, body = the release digest: theme line, `## Changes`, `## Gates`, changelog link last); `release-pr-review` reviews and fixes on that branch (each fix an ordinary commit on top of the stack, pushed plainly — nothing already pushed is ever rewritten, so `main` keeps the record of what the review corrected — PR body kept in sync, one summary comment); then `release-and-publish` fast-forwards `main` locally with `git merge --ff-only`, creates the tag on `main`'s tip, pushes `main` and the tag, deletes the branch, and publishes. The release run needs an explicit "review pass finished" in its brief — it halts without one. **Never merge through the GitHub UI or `gh pr merge`**: squash and rebase-merge are disabled in the repo settings because both rewrite the stack (rebase-merge also strips the SSH signatures), and a merge commit breaks the linear history.
360
369
 
361
370
  ---
362
371
 
@@ -388,7 +397,7 @@ import { getEuropePmcService } from '@/services/europe-pmc/europe-pmc-service.js
388
397
  - [ ] `BIORXIV_MAILTO` included in outbound User-Agent header when set (optional env var)
389
398
  - [ ] DOI input normalized and validated with `normalizeDoi()` before calling API
390
399
  - [ ] Two-server fan-out uses `Promise.allSettled`; partial failures reported per-DOI in `failed[]`
391
- - [ ] `biorxiv_list_recent` with `server="both"` surfaces per-server pagination state, not a merged cursor
400
+ - [ ] `biorxiv_list_recent` and `biorxiv_list_published_versions` with `server="both"` surface per-server pagination state, not a merged cursor
392
401
  - [ ] Registered in `createApp()` arrays (directly or via barrel exports)
393
402
  - [ ] Tests use `createMockContext()` from `@cyanheads/mcp-ts-core/testing`
394
403
  - [ ] `.codex-plugin/plugin.json` populated — `name`, `version`, `description`, `repository`, `license` from `package.json`; `interface.displayName` = the unscoped repo name (never the npm scope — `lint:packaging` enforces this); `interface.shortDescription` from `package.json` description
package/CLAUDE.md CHANGED
@@ -1,11 +1,11 @@
1
1
  # Agent Protocol
2
2
 
3
3
  **Server:** biorxiv-mcp-server
4
- **Version:** 0.3.0
5
- **Framework:** [@cyanheads/mcp-ts-core](https://www.npmjs.com/package/@cyanheads/mcp-ts-core) `^0.13.6`
4
+ **Version:** 0.4.0
5
+ **Framework:** [@cyanheads/mcp-ts-core](https://www.npmjs.com/package/@cyanheads/mcp-ts-core) `^0.13.15`
6
6
  **Engines:** Bun ≥1.4.0, Node ≥24.0.0
7
- **MCP SDK:** `@modelcontextprotocol/server` ^2.0.0
8
- **Zod:** ^4.6.5
7
+ **MCP SDK:** `@modelcontextprotocol/server` ^2.3.1
8
+ **Zod:** ^4.6.6
9
9
 
10
10
  > **Read the framework docs first:** `node_modules/@cyanheads/mcp-ts-core/CLAUDE.md` contains the full API reference — builders, Context, error codes, exports, patterns. This file covers server-specific conventions only.
11
11
 
@@ -105,8 +105,18 @@ export function getServerConfig(): z.infer<typeof ServerConfigSchema> {
105
105
 
106
106
  `parseEnvConfig` maps Zod schema paths → env var names so validation errors name the actual variable rather than the internal path. It reads an empty string and an unsubstituted whole-value `${…}` placeholder (an MCPB optional field left blank) as unset, so an optional field stays `undefined` and a defaulted field takes its default — no per-field `z.preprocess` guard needed.
107
107
 
108
+ For env booleans use `z.stringbool()`, never `z.coerce.boolean()` — `Boolean("false")` is `true`, so a coerced flag can't be disabled through the environment. `z.stringbool()` parses `true/false/1/0/yes/no/on/off` and rejects anything else, so `=false` actually disables.
109
+
110
+ ### Server identity and instructions
111
+
112
+ `createApp()` accepts optional identity fields forwarded to the SDK's `initialize` response and the server manifest (`/.well-known/mcp.json`): `name`, `title`, `websiteUrl`, `description` (wins over `MCP_SERVER_DESCRIPTION`), `icons`, and `instructions`. This server sets `name` and `title` (both the repo name) and `instructions` in `src/index.ts`.
113
+
114
+ `instructions` is optional server-level orientation, sent on every `initialize` as session-level context. Use it for deployment guidance (connection aliases, regional notes, scope hints) instead of repeating the same context across tool descriptions. Client adoption is uneven, but there's no downside when set.
115
+
108
116
  ### Session posture and shutdown
109
117
 
118
+ Two more `createApp()` options shape how the server runs rather than how it presents itself:
119
+
110
120
  ```ts
111
121
  await createApp({
112
122
  sessionMode: 'stateless', // or { default: 'stateful', require: 'stateful' }
@@ -115,9 +125,9 @@ await createApp({
115
125
  });
116
126
  ```
117
127
 
118
- `sessionMode` declares the HTTP session posture in `src/` instead of leaving it to a deployment's `MCP_SESSION_MODE`, which still wins whenever it carries a meaningful value. This server declares `stateless` — no tool calls `ctx.requestInput`. Add `require: 'stateful'` if a tool ever asks the caller for input mid-handler: startup then fails with a `ConfigurationError` rather than serving a mode in which a 2025-era client can never answer the prompt. Stdio is never refused.
128
+ `sessionMode` declares the HTTP session posture in `src/` instead of leaving it to a deployment's `MCP_SESSION_MODE`, which still wins whenever it carries a meaningful value (an empty string and an unsubstituted `${…}` placeholder read as unset and fall through to the option). This server declares `stateless` — no tool calls `ctx.requestInput`. Add `require: 'stateful'` if a tool ever asks the caller for input mid-handler: startup then fails with a `ConfigurationError` rather than serving a mode in which a 2025-era client can never answer the prompt. Stdio is never refused.
119
129
 
120
- `teardown(core)` is the `setup()` counterpart — release a watcher, socket, or non-`unref()`'d timer there. It runs after the transport stops and before the logger closes, on every shutdown path. This server's services allocate none, so it declares no `teardown`.
130
+ `teardown(core)` is the `setup()` counterpart — release a watcher, socket, or non-`unref()`'d timer there. It runs after the transport stops and before the logger closes, on every shutdown path, and a signal-triggered shutdown then exits the process explicitly (0, or 1 if a step never settles within the framework's 10 s ceiling). This server's services allocate none, so it declares no `teardown`.
121
131
 
122
132
  ---
123
133
 
@@ -128,14 +138,14 @@ Handlers receive a unified `ctx` object. Key properties:
128
138
  | Property | Description |
129
139
  |:---------|:------------|
130
140
  | `ctx.log` | Request-scoped logger — `.debug()`, `.info()`, `.notice()`, `.warning()`, `.error()`. Auto-correlates requestId, traceId, tenantId. Dual-sink: Pino **and** `notifications/message` to the client, so treat it as client-visible. |
131
- | `ctx.state` | Tenant-scoped KV — `.get(key)`, `.set(key, value, { ttl? })`, `.delete(key)`, `.getMany(keys)`, `.list(prefix, { cursor, limit })`. Accepts any serializable value. |
141
+ | `ctx.state` | Tenant-scoped KV — `.get(key)`, `.set(key, value, { ttl? })`, `.delete(key)`, `.getMany(keys)`, `.list(prefix, { cursor, limit })`. Accepts any JSON-serializable value; reads return its JSON form (a `Date` comes back as an ISO string). |
132
142
  | `ctx.requestInput` | Suspend and ask the caller for more input — `return ctx.requestInput({ inputRequests: { key: inputRequired.elicit({ message, requestedSchema }) } })`. Never returns; the handler is re-entered with the answers. Always present. |
133
- | `ctx.inputs` | Reader over a retried request's responses — `.accepted(key, schema)`, `.view(key)`, `.state()`, `.dropped`. Empty on the first round. |
134
- | `ctx.recoveryFor(reason)` | Typed lookup of the contract `recovery` for a declared reason. Returns `{ recovery: { hint } }` for known reasons, `{}` otherwise. Spread into `ctx.fail` data to mirror the contract hint into `content[]`. |
143
+ | `ctx.inputs` | The request's responses — `.accepted(key, schema)`, `.view(key)`, `.state()`, `.dropped` — limited to what the client declared (`elicitation` and its form/url modes, `sampling`, `roots`). Client-supplied: a consent gate trusts only a `ctx.state` record it stored when it asked, bound to the operation, caller, and target (see the `api-context` skill). |
144
+ | `ctx.clientCapabilities` | What the client declared for this request, `undefined` when no view exists. Decides whether to ask for optional context (e.g. roots); never a reason to skip a consent prompt. |
135
145
  | `ctx.enrich` | Success-path agent context (empty-result notices, query echo, pagination totals) — `ctx.enrich(...)` or `.notice()` / `.total()` / `.echo()` / `.truncated()`. Reaches `structuredContent` and `content[]`; lands only when the definition declares an `enrichment` block (no-op otherwise). |
136
146
  | `ctx.content` | Non-text content blocks — `.image(data, mimeType)`, `.audio(data, mimeType)`, or `ctx.content(block)` for a raw block. Prepended to `content[]` after `format()`; never enters `structuredContent`. |
137
147
  | `ctx.signal` | `AbortSignal` for cancellation. |
138
- | `ctx.requestId` | Unique request ID. |
148
+ | `ctx.requestId` | Request ID — the one every log record of the call carries and its error envelope returns as `data.requestId`. |
139
149
  | `ctx.tenantId` | Tenant ID from JWT; `'default'` for stdio or HTTP with auth off. |
140
150
 
141
151
  ---
@@ -144,7 +154,7 @@ Handlers receive a unified `ctx` object. Key properties:
144
154
 
145
155
  Handlers throw — the framework catches, classifies, and formats.
146
156
 
147
- **Recommended: typed error contract.** Declare `errors: [{ reason, code, when, recovery, retryable?, severity?, thrownBy? }]` on `tool()` / `resource()` to receive a typed `ctx.fail(reason, …)` keyed by the declared reason union. TypeScript catches `ctx.fail('typo')` at compile time, `data.reason` is auto-populated for observability, and the linter enforces conformance against the handler body. The `recovery` field is required (≥ 5 words, lint-validated). Spread `ctx.recoveryFor('reason')` into `data` to mirror the contract recovery onto the wire (`data.recovery.hint`, mirrored into `content[]` text unless the message already contains it verbatim), or pass an explicit `recovery: { hint }` when the hint needs runtime context (a computed wait). Forwarding is lint-enforced per throw site — a `ctx.fail` carrying neither warns as `error-contract-recovery-unforwarded`. Every declared reason must be named by a literal `ctx.fail('<reason>'` or `ctx.recoveryFor('<reason>')` in its handler — `lint:mcp`'s `error-contract-unthrown` flags one that isn't; mark an entry the service layer throws with `thrownBy: 'service'` (lint-only metadata) instead. Baseline codes (`InternalError`, `ServiceUnavailable`, `Timeout`, `ValidationError`, `SerializationError`, `RequestCancelled`) bubble freely and don't need declaring.
157
+ **Recommended: typed error contract.** Declare `errors: [{ reason, code, when, recovery, retryable?, severity?, thrownBy? }]` on `tool()` / `resource()` to receive `ctx.fail(reason, …)` typed against the reason union. TypeScript catches typos at compile time, `data.reason` is auto-populated for observability, linter enforces conformance against the handler body. `recovery` is required (≥ 5 words, lint-validated) — the single source of truth for the agent's next move. The framework puts it on the wire whenever a failure carrying that `reason` arrives without a hint — a bare `ctx.fail('reason')` or a service throw with `data: { reason }` — as `data.recovery.hint`, mirrored into `content[]` text unless the message already contains it verbatim; override with an explicit `{ recovery: { hint: '...' } }` when dynamic runtime context matters (a computed wait). Every error envelope also carries `data.requestId`, the id the server's log records for that call carry, and `content[]` closes with `(reason … · request <id>)`. Every declared reason must be named by a literal `ctx.fail('<reason>'` or `ctx.recoveryFor('<reason>')` in its handler — `lint:mcp`'s `error-contract-unthrown` flags one that isn't; mark an entry the service layer throws with `thrownBy: 'service'` so the rule skips it — lint-only metadata, nothing at runtime reads it. Baseline codes (`InternalError`, `ServiceUnavailable`, `Timeout`, `ValidationError`, `SerializationError`, `RequestCancelled`) bubble freely and don't need declaring.
148
158
 
149
159
  ```ts
150
160
  import { JsonRpcErrorCode } from '@cyanheads/mcp-ts-core/errors';
@@ -157,9 +167,7 @@ errors: [
157
167
  async handler(input, ctx) {
158
168
  const result = await getBiorxivApiService().getDetails(input.doi, input.server ?? 'both');
159
169
  if (result.collection.length === 0) {
160
- throw ctx.fail('doi_not_found', `DOI not found: ${input.doi}`, {
161
- ...ctx.recoveryFor('doi_not_found'),
162
- });
170
+ throw ctx.fail('doi_not_found', `DOI not found: ${input.doi}`);
163
171
  }
164
172
  return result;
165
173
  }
@@ -167,7 +175,7 @@ async handler(input, ctx) {
167
175
 
168
176
  **Declare contracts inline on each tool.** The contract is part of the tool's public surface — one file should give the full picture. Don't extract a shared `errors[]` constant; per-tool repetition is the intended cost of locality.
169
177
 
170
- **Fallback (no contract entry fits):** error factories or plain `Error`.
178
+ **Fallback (no contract entry fits):** throw via factories or plain `Error`.
171
179
 
172
180
  ```ts
173
181
  import { notFound, serviceUnavailable } from '@cyanheads/mcp-ts-core/errors';
@@ -175,7 +183,7 @@ throw notFound('Preprint not found', { doi });
175
183
  throw serviceUnavailable('api.biorxiv.org unreachable', { url }, { cause: err });
176
184
  ```
177
185
 
178
- See framework CLAUDE.md and the `api-errors` skill for the full auto-classification table, all factories, and the contract reference.
186
+ See framework CLAUDE.md and the `api-errors` skill for the full auto-classification table, all available factories, and the contract reference.
179
187
 
180
188
  ---
181
189
 
@@ -205,6 +213,7 @@ src/
205
213
  biorxiv-list-recent.tool.ts
206
214
  biorxiv-search-preprints.tool.ts
207
215
  biorxiv-get-published-version.tool.ts
216
+ biorxiv-list-published-versions.tool.ts
208
217
  biorxiv-get-fulltext.tool.ts
209
218
  biorxiv-list-categories.tool.ts
210
219
  ```
@@ -228,12 +237,12 @@ src/
228
237
  - **Category spelling:** the listing API filters only on lowercase with `/` and `_` read as a space (`hiv aids`), and answers anything else with its unfiltered listing, echoing `messages[0].category: "all"`. `BiorxivApiService` owns the spelling and flags `categoryIgnored`; never present a flagged page as filtered.
229
238
  - **Funder data:** upstream funder `name`s and ROR `id`s on records are misattributed — expose only `awards` (verbatim, never split), never a funder name or ID. The listing's `funder` filter takes a bare ROR ID (`normalizeRorId()` strips the URL form and checks the checksum), is bioRxiv-only, and `"funder value not found"` (flagged `funderNotFound`) is an error, never an empty page.
230
239
  - **Server parameter:** `"biorxiv" | "medrxiv" | "both"`. Default to `"both"`, including for DOI-resolution tools — both servers issue `10.1101/` DOIs, so the DOI never identifies the server and a single-server default silently misses records held by the other. Fan out via `Promise.allSettled` when `"both"` and name the answering server in the output.
231
- - **Pagination:** integer `cursor` offset (0, 30, 60, …). Page size is fixed at 30 by the API — do not expose a `limit` parameter.
232
- - **Two-server pagination:** when `server="both"`, surface per-server state: `{ biorxiv: { cursor, total }, medrxiv: { cursor, total } }`. A merged cursor is ambiguous.
240
+ - **Pagination:** integer `cursor` offset. Page size is fixed by the API — 30 on `biorxiv_list_recent` (0, 30, 60, …), 100 on `biorxiv_list_published_versions` (0, 100, 200, …) — do not expose a `limit` parameter.
241
+ - **Two-server pagination:** when `server="both"`, surface per-server state: `{ biorxiv: { cursor, total }, medrxiv: { cursor, total } }`. A merged cursor is ambiguous. A server whose cursor overshot is flagged `exhausted`: the posting-date listing reports `total: 0` there, and the publication-date listing reports no total at all, so its exhausted entry carries none.
233
242
  - **EuropePMC enrichment:** `biorxiv_search_preprints` uses EuropePMC for relevance then enriches matching DOIs via the details endpoint. Enrichment routes based on `input.server` — when explicit ("biorxiv" or "medrxiv"), always use that server. `10.1101/` prefix is used as a hint only within `server="both"` to prefer bioRxiv first, but is not a reliable discriminator (both servers share this prefix). The ranked search stays `resulttype=lite`, which never carries an abstract; records that fall back to EuropePMC metadata get theirs from one DOI-keyed `resultType=core` query (`getAbstracts`), made only when abstracts are requested and something fell back. A failed lookup is a notice, never a failed search.
234
- - **`format()` must be content-complete:** Claude Desktop reads `content[]` from `format()`, not `structuredContent`. Revision list, crosswalk data, pagination state, and every abstract a record carries must all appear in the rendered markdown, not just counts. `include_abstract` (default `false` on `biorxiv_list_recent`, `true` on `biorxiv_search_preprints`) removes the abstract from both surfaces together and no other field.
243
+ - **`format()` must be content-complete:** Claude Desktop reads `content[]` from `format()`, not `structuredContent`. Revision list, crosswalk data, pagination state, and every abstract a record carries must all appear in the rendered markdown, not just counts. `include_abstract` (default `false` on `biorxiv_list_recent` and `biorxiv_list_published_versions`, `true` on `biorxiv_search_preprints`) removes the abstract from both surfaces together and no other field.
235
244
  - **Upstream text in `format()`:** run titles and abstracts through `normalizeUpstreamText()` in the service (never in `format()`), and pass every upstream free-text field through `escapeMarkdown()` where `format()` interpolates it. It escapes only characters that would change rendering (`A\*T`, `\<COMP id>`; `p<0.05` and `cc_by` stay raw), because agents copy spans out of `content[]` — keep it that narrow. Never escape DOIs, URLs, `structuredContent`, or `biorxiv_get_fulltext`'s extracted `content`.
236
- - **Polite access:** include `BIORXIV_MAILTO` in the `User-Agent` header when set: `biorxiv-mcp-server/0.1.0 (mailto:${config.mailto})`. The env var is optional — omit the mailto segment when not configured.
245
+ - **Polite access:** include `BIORXIV_MAILTO` in the `User-Agent` header when set: `biorxiv-mcp-server/${SERVER_VERSION} (mailto:${config.mailto})`. The env var is optional — omit the mailto segment when not configured.
237
246
 
238
247
  ---
239
248
 
@@ -265,21 +274,21 @@ Available skills:
265
274
  | `release-and-publish` | Fast-forward merge (release PR mode) + tag + push + npm + MCP Registry + GH Release + Docker. Picks up from `git-wrapup` |
266
275
  | `maintenance` | Investigate changelogs, adopt upstream changes, sync skills to agent dirs |
267
276
  | `orchestrations` | Chain task skills into a gated multi-phase pipeline — build-out, QA-fix, update-ship — when you can spawn sub-agents |
277
+ | `report-issue-framework` | File a bug or feature request against `@cyanheads/mcp-ts-core` via `gh` CLI |
278
+ | `report-issue-local` | File a bug or feature request against this server's own repo via `gh` CLI |
279
+ | `techniques` | Catalog of response/data-shaping techniques — overflow handling, payload shaping, retrieval patterns |
268
280
  | `api-auth` | Auth modes, scopes, JWT/OAuth |
269
281
  | `api-canvas` | DataCanvas: register tabular data, run SQL, export, plus the `spillover()` helper for big result sets — Tier 3 opt-in |
270
282
  | `api-config` | AppConfig, parseConfig, env vars |
271
283
  | `api-context` | Context interface, RequestContext, logger, state, multi-round-trip input |
272
284
  | `api-errors` | McpError, JsonRpcErrorCode, error patterns |
273
285
  | `api-linter` | Definition linter rule catalog — invoked by `bun run lint:mcp` and `devcheck` |
274
- | `api-mirror` | MirrorService: persistent, self-refreshing local mirror of a bulk upstream dataset (embedded SQLite + FTS5) — Tier 3 opt-in |
286
+ | `api-mirror` | MirrorService: persistent self-refreshing local mirror (embedded SQLite + FTS5) of a bulk upstream dataset — Tier 3 opt-in |
275
287
  | `api-services` | LLM, Speech, Graph services |
276
- | `api-telemetry` | OTel catalog: spans, metrics, completion logs, env config, cardinality rules |
277
288
  | `api-testing` | createMockContext, test patterns |
278
289
  | `api-utils` | Formatting, parsing, security, pagination, scheduling, telemetry helpers |
290
+ | `api-telemetry` | OTel catalog: spans, metrics, completion logs, env config, cardinality rules |
279
291
  | `api-workers` | Cloudflare Workers runtime |
280
- | `report-issue-framework` | File a bug or feature request against `@cyanheads/mcp-ts-core` via `gh` CLI |
281
- | `report-issue-local` | File a bug or feature request against this server's own repo via `gh` CLI |
282
- | `techniques` | Catalog of response/data-shaping techniques — overflow handling, payload shaping, retrieval patterns |
283
292
 
284
293
  **Chaining skills into pipelines.** When the user wants a multi-phase effort — build this server out, QA-and-fix the surface, update-and-ship — *and you can spawn sub-agents*, `framework-skills/orchestrations/SKILL.md` sequences the task skills above into a gated pipeline with verification at each step. Read it to drive the run. Optional: skip it if you can't orchestrate sub-agents, and ignore it entirely if you were *spawned* as one — you've already been scoped to a single phase.
285
294
 
@@ -299,18 +308,18 @@ When you complete a skill's checklist, check the boxes and add a completion time
299
308
  | `bun run devcheck` | Lint + format + typecheck + security + changelog sync |
300
309
  | `bun run audit:fix` | `bun audit fix` — upgrade vulnerable packages to the lowest safe version within existing ranges (`--dry-run` previews, `--latest` rewrites ranges). First response when `devcheck` flags a transitive advisory; then `bun update <name>`, then `bun dedupe` |
301
310
  | `bun run audit:refresh` | Delete `bun.lock` and reinstall. Last resort after `audit:fix`, `bun update <name>`, and `bun dedupe` — re-resolves every ranged dep (the framework pin included) and rewrites the lockfile as `lockfileVersion: 2` |
311
+ | `bun run lint:mcp` | Run the MCP definition linter standalone (rule catalog: `api-linter` skill) |
312
+ | `bun run lint:packaging` | Packaging surface checks — `server.json`/`manifest.json` env-var parity (run by devcheck) |
313
+ | `bun run list-skills` | Print the skill registry |
302
314
  | `bun run tree` | Generate directory structure doc |
303
315
  | `bun run format` | Auto-fix formatting (safe fixes only) |
304
316
  | `bun run format:unsafe` | Also apply Biome's unsafe autofixes — review the diff; they can change behavior |
305
317
  | `bun run test` | Run tests (Vitest — use `bun run test`, not `bun test`) |
306
- | `bun run lint:mcp` | Run the MCP definition linter standalone (rule catalog: `api-linter` skill) |
307
- | `bun run lint:packaging` | Packaging surface checks — `server.json`/`manifest.json` env-var parity (run by devcheck) |
308
- | `bun run bundle` | Build, pack, and clean a `.mcpb` for one-click Claude Desktop install |
309
- | `bun run list-skills` | Surface available local skills for sub-agents |
310
318
  | `bun run start:stdio` | Production mode (stdio) |
311
319
  | `bun run start:http` | Production mode (HTTP) |
312
320
  | `bun run changelog:build` | Regenerate `CHANGELOG.md` from `changelog/*.md` |
313
321
  | `bun run changelog:check` | Verify `CHANGELOG.md` is in sync (used by devcheck) |
322
+ | `bun run bundle` | Build, pack, and clean a `.mcpb` for one-click Claude Desktop install |
314
323
  | `bun run release:github` | Create (or repair) a GitHub Release on the current version's annotated tag |
315
324
 
316
325
  **CI is one file.** `.github/workflows/codeql.yml` is the only GitHub Actions workflow: CodeQL is GitHub-owned end to end, and the file runs only while the repo's CodeQL *default setup* is turned off. Verification — `devcheck`, tests, the release gates — runs locally; don't add a workflow that re-runs it.
@@ -356,7 +365,7 @@ security: false # optional — true ONLY for a source
356
365
 
357
366
  ## Publishing
358
367
 
359
- **Every release goes through a gated release PR** — `git-wrapup`'s "Release PR mode", mode `gated`. Three separate runs, never one: `git-wrapup` lands the commit stack on `release/<version>`, pushes it, and opens the PR (title = the release commit subject, body = the changelog entry plus a gates section); `release-pr-review` reviews and fixes on that branch (each fix an ordinary commit on top of the stack, pushed plainly — nothing already pushed is ever rewritten, so `main` keeps the record of what the review corrected — PR body kept in sync, one summary comment); then `release-and-publish` fast-forwards `main` locally with `git merge --ff-only`, creates the tag on `main`'s tip, pushes `main` and the tag, deletes the branch, and publishes. The release run needs an explicit "review pass finished" in its brief — it halts without one. **Never merge through the GitHub UI or `gh pr merge`**: squash and rebase-merge are disabled in the repo settings because both rewrite the stack (rebase-merge also strips the SSH signatures), and a merge commit breaks the linear history.
368
+ **Every release goes through a gated release PR** — `git-wrapup`'s "Release PR mode", mode `gated`. Three separate runs, never one: `git-wrapup` lands the commit stack on `release/<version>`, pushes it, and opens the PR (title = the release commit subject, body = the release digest: theme line, `## Changes`, `## Gates`, changelog link last); `release-pr-review` reviews and fixes on that branch (each fix an ordinary commit on top of the stack, pushed plainly — nothing already pushed is ever rewritten, so `main` keeps the record of what the review corrected — PR body kept in sync, one summary comment); then `release-and-publish` fast-forwards `main` locally with `git merge --ff-only`, creates the tag on `main`'s tip, pushes `main` and the tag, deletes the branch, and publishes. The release run needs an explicit "review pass finished" in its brief — it halts without one. **Never merge through the GitHub UI or `gh pr merge`**: squash and rebase-merge are disabled in the repo settings because both rewrite the stack (rebase-merge also strips the SSH signatures), and a merge commit breaks the linear history.
360
369
 
361
370
  ---
362
371
 
@@ -388,7 +397,7 @@ import { getEuropePmcService } from '@/services/europe-pmc/europe-pmc-service.js
388
397
  - [ ] `BIORXIV_MAILTO` included in outbound User-Agent header when set (optional env var)
389
398
  - [ ] DOI input normalized and validated with `normalizeDoi()` before calling API
390
399
  - [ ] Two-server fan-out uses `Promise.allSettled`; partial failures reported per-DOI in `failed[]`
391
- - [ ] `biorxiv_list_recent` with `server="both"` surfaces per-server pagination state, not a merged cursor
400
+ - [ ] `biorxiv_list_recent` and `biorxiv_list_published_versions` with `server="both"` surface per-server pagination state, not a merged cursor
392
401
  - [ ] Registered in `createApp()` arrays (directly or via barrel exports)
393
402
  - [ ] Tests use `createMockContext()` from `@cyanheads/mcp-ts-core/testing`
394
403
  - [ ] `.codex-plugin/plugin.json` populated — `name`, `version`, `description`, `repository`, `license` from `package.json`; `interface.displayName` = the unscoped repo name (never the npm scope — `lint:packaging` enforces this); `interface.shortDescription` from `package.json` description
package/Dockerfile CHANGED
@@ -5,16 +5,16 @@
5
5
  # source code into JavaScript, and prepares the production assets.
6
6
  #
7
7
  # Pinned to $BUILDPLATFORM rather than the target platform: `bun run build` emits
8
- # JavaScript, and only `dist/` crosses into the production stage, which runs its
9
- # own target-arch install. Built for the target instead, the non-native leg of a
10
- # `--platform linux/amd64,linux/arm64` build runs under QEMU, where bun >= 1.4
11
- # aborts with a JavaScriptCore allocator assertion and fails the multi-arch push.
8
+ # JavaScript, and only `dist/` crosses into the production stage. Built for the
9
+ # target instead, the non-native leg of a `--platform linux/amd64,linux/arm64`
10
+ # build runs under QEMU, where bun >= 1.4 aborts with a JavaScriptCore allocator
11
+ # assertion and fails the multi-arch push.
12
12
  #
13
13
  # The constraint this assumes: the build stage produces platform-independent
14
14
  # output. A stage that compiles a native addon needs the target-arch toolchain
15
15
  # and cannot cross-compile this way — drop the flag there.
16
16
  # ==============================================================================
17
- FROM --platform=$BUILDPLATFORM oven/bun:1.4.0 AS build
17
+ FROM --platform=$BUILDPLATFORM oven/bun:1.4.3 AS build
18
18
 
19
19
  WORKDIR /usr/src/app
20
20
 
@@ -34,59 +34,105 @@ RUN bun run build
34
34
 
35
35
 
36
36
  # ==============================================================================
37
- # Production Stage
37
+ # Production Dependencies Stage
38
38
  #
39
- # This stage creates a minimal, optimized, and secure image for running the
40
- # application. It uses a slim base image and only includes production
41
- # dependencies and build artifacts.
39
+ # Installs the production dependency tree for the target platform. Every step
40
+ # here can run JavaScript — bunfig.toml's security scanner runs as a Bun
41
+ # program, and so do the OTel and musl-prune scripts — so the stage runs on
42
+ # $BUILDPLATFORM and cross-installs with `--os`/`--cpu`, which pick each
43
+ # platform-specific optional dependency (native bindings) for the target. Only
44
+ # `node_modules` leaves this stage.
45
+ #
46
+ # A clean image rather than `FROM build`: the build stage's node_modules holds
47
+ # devDependencies.
42
48
  # ==============================================================================
43
- FROM oven/bun:1.4.0-slim AS production
49
+ FROM --platform=$BUILDPLATFORM oven/bun:1.4.3 AS deps
44
50
 
45
51
  WORKDIR /usr/src/app
46
52
 
47
- # Set the environment to production for performance and to ensure only
48
- # production dependencies are installed.
49
- ENV NODE_ENV=production
50
-
51
- # OCI image metadata (https://github.com/opencontainers/image-spec/blob/main/annotations.md)
52
- ARG APP_VERSION
53
- LABEL org.opencontainers.image.title="@cyanheads/biorxiv-mcp-server"
54
- LABEL org.opencontainers.image.description="Search and retrieve bioRxiv and medRxiv preprints — by DOI, date interval, or keyword — via MCP. STDIO or Streamable HTTP."
55
- LABEL org.opencontainers.image.licenses="Apache-2.0"
56
- LABEL org.opencontainers.image.version="${APP_VERSION}"
57
- LABEL org.opencontainers.image.source="https://github.com/cyanheads/biorxiv-mcp-server"
58
-
59
- # Copy dependency manifests
60
- COPY package.json bun.lock ./
53
+ # Copy dependency manifests. `bunfig.toml` rides along so every install below
54
+ # passes its release-age gate and security scanner, as a local install does.
55
+ COPY package.json bun.lock bunfig.toml ./
56
+
57
+ # The scanner bunfig.toml names is a devDependency, and Bun installs a missing
58
+ # scanner through the same production-filtered install, which omits it and
59
+ # aborts. Seed it from the build stage's full install instead. Remove this line,
60
+ # and the `rm` at the end of this stage, if bunfig.toml stops naming a scanner.
61
+ COPY --from=build /usr/src/app/node_modules/@socketsecurity/bun-security-scanner ./node_modules/@socketsecurity/bun-security-scanner
62
+
63
+ # Docker names the target architecture `amd64`/`arm64`; Bun's `--cpu` takes
64
+ # `x64`/`arm64`. Mapped once here, read by both installs below. `oven/bun`
65
+ # publishes only these two architectures, so any other target fails here.
66
+ ARG TARGETOS
67
+ ARG TARGETARCH
68
+ RUN case "$TARGETARCH" in \
69
+ amd64) echo x64 ;; \
70
+ arm64) echo arm64 ;; \
71
+ *) echo "Unsupported TARGETARCH '$TARGETARCH': expected amd64 or arm64" >&2; exit 1 ;; \
72
+ esac > .bun-cpu
61
73
 
62
74
  # Install only production dependencies, ignoring any lifecycle scripts (like 'prepare')
63
75
  # that are not needed in the final production image.
64
76
  # `--omit=peer` drops the framework's optional peer tiers (test runner, service
65
77
  # SDKs, parsers) that Bun would otherwise auto-install. Anything this server
66
78
  # actually imports belongs in its own `dependencies`, so nothing needed at
67
- # runtime is lost. The OTEL step below carries the same flag — without it, that
68
- # install re-resolves the graph and pulls every optional peer back in.
79
+ # runtime is lost.
69
80
  RUN --mount=type=cache,target=/root/.bun/install/cache \
70
- bun install --production --omit=peer --frozen-lockfile --ignore-scripts
81
+ bun install --production --omit=peer --frozen-lockfile --ignore-scripts \
82
+ --os="$TARGETOS" --cpu="$(cat .bun-cpu)"
71
83
 
72
84
  # Conditionally install OpenTelemetry optional peer dependencies (Tier 3).
73
- # These are not bundled by default to keep the base image lean. Enable at build time
74
- # with: docker build --build-arg OTEL_ENABLED=true
85
+ # Installed by default. Omit them for a leaner image at build time
86
+ # with: docker build --build-arg OTEL_ENABLED=false
87
+ # The script reads the list and each range from the installed framework's
88
+ # `peerDependencies` and passes the target flags on to its `bun install`.
89
+ COPY scripts/install-otel.ts ./scripts/
75
90
  ARG OTEL_ENABLED=true
76
91
  RUN --mount=type=cache,target=/root/.bun/install/cache \
77
92
  if [ "$OTEL_ENABLED" = "true" ]; then \
78
- bun add --omit=dev --omit=peer --ignore-scripts @hono/otel \
79
- @opentelemetry/instrumentation-http \
80
- @opentelemetry/exporter-metrics-otlp-http \
81
- @opentelemetry/exporter-trace-otlp-http \
82
- @opentelemetry/instrumentation-pino \
83
- @opentelemetry/resources \
84
- @opentelemetry/sdk-metrics \
85
- @opentelemetry/sdk-node \
86
- @opentelemetry/sdk-trace-node \
87
- @opentelemetry/semantic-conventions; \
93
+ bun scripts/install-otel.ts --os="$TARGETOS" --cpu="$(cat .bun-cpu)"; \
88
94
  fi
89
95
 
96
+ # `--os`/`--cpu` have no libc counterpart, so a native dependency published in
97
+ # glibc and musl variants installs both. The runtime image is Debian (glibc)
98
+ # and never loads the musl copy; the script deletes every package whose own
99
+ # `libc` admits only musl. It follows every install, since a later
100
+ # `bun install` restores what it removes.
101
+ COPY scripts/prune-musl-packages.ts ./scripts/
102
+ RUN bun scripts/prune-musl-packages.ts
103
+
104
+ # The seeded scanner served only the installs above; keep it out of the image.
105
+ RUN rm -rf node_modules/@socketsecurity/bun-security-scanner
106
+
107
+
108
+ # ==============================================================================
109
+ # Production Stage
110
+ #
111
+ # This stage creates a minimal, optimized, and secure image for running the
112
+ # application. It uses a slim base image and only includes production
113
+ # dependencies and build artifacts. Its only Bun invocations are HEALTHCHECK
114
+ # and CMD, which run on the real target at container start.
115
+ # ==============================================================================
116
+ FROM oven/bun:1.4.3-slim AS production
117
+
118
+ WORKDIR /usr/src/app
119
+
120
+ # Set the environment to production for performance.
121
+ ENV NODE_ENV=production
122
+
123
+ # OCI image metadata (https://github.com/opencontainers/image-spec/blob/main/annotations.md)
124
+ ARG APP_VERSION
125
+ LABEL org.opencontainers.image.title="@cyanheads/biorxiv-mcp-server"
126
+ LABEL org.opencontainers.image.description="Search and retrieve bioRxiv and medRxiv preprints — by DOI, date interval, or keyword — via MCP. STDIO or Streamable HTTP."
127
+ LABEL org.opencontainers.image.licenses="Apache-2.0"
128
+ LABEL org.opencontainers.image.version="${APP_VERSION}"
129
+ LABEL org.opencontainers.image.source="https://github.com/cyanheads/biorxiv-mcp-server"
130
+
131
+ # The manifest comes from the build context: the deps stage's copy was rewritten
132
+ # by the OTel install, and the runtime reads only its name, version, and type.
133
+ COPY package.json ./
134
+ COPY --from=deps /usr/src/app/node_modules ./node_modules
135
+
90
136
  # Copy the compiled application code from the build stage
91
137
  COPY --from=build /usr/src/app/dist ./dist
92
138
 
@@ -111,7 +157,6 @@ ENV MCP_TRANSPORT_TYPE="http"
111
157
  ENV MCP_SESSION_MODE="stateless"
112
158
  ENV MCP_LOG_LEVEL="info"
113
159
  ENV LOGS_DIR="/var/log/biorxiv-mcp-server"
114
- ENV MCP_FORCE_CONSOLE_LOGGING="true"
115
160
 
116
161
  # Expose the port the server listens on
117
162
  EXPOSE ${MCP_HTTP_PORT}