prospectapis-mcp 0.2.1 → 0.2.3

package/CHANGELOG.md

@@ -6,6 +6,21 @@ All notable changes to `prospectapis-mcp` are documented here. The format follow
 
 ## [Unreleased]
 
+## [0.2.3] - 2026-10-11
+
+### Added
+
+- `server.json` (the Official MCP Registry manifest) and the matching `mcpName` field in `package.json`, both `io.github.ProspectAPIs/prospectapis-mcp`, so the server can be listed on registry.modelcontextprotocol.io (which PulseMCP ingests). No tool or behaviour change.
+- `scripts/registry-drift.mjs` with `npm run check:registry-drift` and a `postpublish` hook, so every npm release also updates the registry entry and a drift between npm, `server.json` and the registry fails the check.
+
+## [0.2.2] - 2026-10-03
+
+### Changed
+
+- `research_submit` and `research_get` describe the brief's new `social` section: the company's X account and Reddit mentions, each value cited, absent when nothing was found.
+
+- Tests that compare the tool schemas with the API source moved to `test/api-parity.test.js`, which runs only where that source is present; the rest of the suite runs in a standalone checkout. `npm publish` now runs the tests first (`prepublishOnly`).
+
 ## [0.2.1] - 2026-10-03
 
 ### Added

package/package.json

@@ -1,6 +1,7 @@
 {
   "name": "prospectapis-mcp",
-  "version": "0.2.1",
+  "version": "0.2.3",
+  "mcpName": "io.github.ProspectAPIs/prospectapis-mcp",
   "description": "MCP server for ProspectAPIs: search funding rounds, run paid prospect research briefs, manage funding watchlists, read your balance and the live price list.",
   "type": "module",
   "bin": {
@@ -24,7 +25,10 @@
   "scripts": {
     "start": "node src/index.js",
     "check": "node --check src/index.js && node --check src/server.js && node --check src/tools.js",
-    "test": "node --test \"test/**/*.test.js\""
+    "test": "node --test \"test/**/*.test.js\"",
+    "prepublishOnly": "npm test && node scripts/prepublish-tenant-check.mjs",
+    "check:registry-drift": "node scripts/registry-drift.mjs",
+    "postpublish": "node scripts/registry-drift.mjs --publish"
   },
   "dependencies": {
     "@modelcontextprotocol/sdk": "^1.29.0",

package/src/tools.js

@@ -78,6 +78,7 @@ export const TOOLS = [
     body: true,
     description:
       "Start a cited pre-call research brief on one prospect: company, funding, traction, competitors, news, people, risks, regions, " +
+      "social (their X account and Reddit mentions, when found), " +
       "plus a `fit` section when you pass `context`. Give exactly ONE identifier (domain, website_url, company_name, company_linkedin_url, " +
       "a work email, or linkedin_url), or person_name together with company_name. Returns `research_id` at once; the brief takes 20 to 90 seconds, " +
       "so poll research_get every 10 to 15 seconds until `status` is `completed` or `failed`. " +
@@ -105,7 +106,10 @@ export const TOOLS = [
     write: false,
     description:
       "Read a research job by the `research_id` research_submit returned: `status` (queued, running, completed, failed), `input`, " +
-      "`result` (the brief, once completed) and `error` (once failed). Free; only your own account's jobs are visible.",
+      "`result` (the brief, once completed) and `error` (once failed). Every brief field is {value, sources, confidence}. " +
+      "`result.social` holds `x` (handle, followers, bio, posting_cadence, recent_topics, notable_posts) and `reddit` " +
+      "(mention_count_90d, top_threads, communities, tone, tone_basis); a platform with nothing found is absent, and with " +
+      "neither `social` is absent. Free; only your own account's jobs are visible.",
     shape: {
       id: z.string().uuid().describe("The `research_id` from research_submit."),
     },