feat: tag outgoing data-fair API calls with a /mcp referer

src/mcp-servers/datasets/tools.test.ts

@@ -6,20 +6,22 @@ import { Client } from '@modelcontextprotocol/sdk/client/index.js'
 import { InMemoryTransport } from '@modelcontextprotocol/sdk/inMemory.js'
 import nock from 'nock'
 import registerTools from './tools.ts'
+import { buildAxiosOptions } from './tools/_utils.ts'
+import config from '#config'
 
 /**
  * Fake Data Fair API server that returns canned responses based on the URL path.
  * Routes are matched longest-pattern-first to avoid ambiguity.
  */
-const routes: Record<string, (url: URL) => any> = {}
+const routes: Record<string, (url: URL, req: IncomingMessage) => any> = {}
 
 const fakeApi = createServer((req: IncomingMessage, res: ServerResponse) => {
   const url = new URL(req.url!, 'http://localhost')
   // Sort patterns by length descending so more specific routes match first
   const sortedPatterns = Object.keys(routes).sort((a, b) => b.length - a.length)
   for (const pattern of sortedPatterns) {
     if (url.pathname.includes(pattern)) {
-      const body = JSON.stringify(routes[pattern](url))
+      const body = JSON.stringify(routes[pattern](url, req))
       res.writeHead(200, { 'content-type': 'application/json' })
       res.end(body)
       return
@@ -72,6 +74,28 @@ after(async () => {
   await new Promise<void>(resolve => fakeApi.close(() => resolve()))
 })
 
+describe('Referer header on outgoing data-fair calls', () => {
+  // The Referer is only meant to tag traffic when the MCP is self-hosted as part
+  // of the data-fair stack (no portalUrl, origin resolved from X-Forwarded-* headers).
+  // In standalone mode (portalUrl set, remote data-fair) no Referer is sent.
+  it('sets a Referer ending in /mcp when self-hosted (stack mode, no portalUrl)', () => {
+    const saved = config.portalUrl
+    config.portalUrl = undefined
+    try {
+      const headers = buildAxiosOptions('https://data.example.com').headers as Record<string, string>
+      assert.equal(headers.Referer, 'https://data.example.com/mcp')
+    } finally {
+      config.portalUrl = saved
+    }
+  })
+
+  it('omits the Referer in standalone mode (portalUrl set)', () => {
+    // portalUrl is set by the `before` hook above (standalone mode)
+    const headers = buildAxiosOptions('https://data.example.com').headers as Record<string, string>
+    assert.equal(headers.Referer, undefined)
+  })
+})
+
 describe('list_datasets', () => {
   it('should list datasets and return formatted results', async () => {
     routes['/catalog/datasets'] = (url) => ({

src/mcp-servers/datasets/tools/_utils.ts

@@ -54,10 +54,11 @@ export const getOrigin = (headers: IsomorphicHeaders | undefined): string => {
   }
 }
 
-export const buildAxiosOptions = (headers: IsomorphicHeaders | undefined): AxiosRequestConfig => {
+export const buildAxiosOptions = (baseUrl: string): AxiosRequestConfig => {
   const axiosHeaders: Record<string, string> = {
-    'User-Agent': '@data-fair/mcp (Datasets)'
+    'User-Agent': '@data-fair/mcp (Datasets)',
   }
+  if (!config.portalUrl) axiosHeaders['Referer'] = baseUrl + '/mcp'
   if (config.dataFairAPIKey) {
     axiosHeaders['x-api-key'] = config.dataFairAPIKey
   }

src/mcp-servers/datasets/tools/aggregate-data.ts

@@ -56,7 +56,7 @@ export default (server: McpServer) => {
       try {
         response = (await axios.get(
           fetchUrl.toString(),
-          buildAxiosOptions(extra.requestInfo?.headers)
+          buildAxiosOptions(baseUrl)
         )).data
       } catch (err: any) {
         handleApiError(err)

src/mcp-servers/datasets/tools/calculate-metric.ts

@@ -49,7 +49,7 @@ export default (server: McpServer) => {
       try {
         response = (await axios.get(
           fetchUrl.toString(),
-          buildAxiosOptions(extra.requestInfo?.headers)
+          buildAxiosOptions(baseUrl)
         )).data
       } catch (err: any) {
         handleApiError(err)

src/mcp-servers/datasets/tools/describe-dataset.ts

@@ -67,7 +67,7 @@ export default (server: McpServer) => {
       try {
         fetchedData = (await axios.get(
           new URL(`/data-fair/api/v1/datasets/${encodeURIComponent(params.datasetId)}`, baseUrl).toString(),
-          buildAxiosOptions(extra.requestInfo?.headers)
+          buildAxiosOptions(baseUrl)
         )).data
       } catch (err: any) {
         handleApiError(err)
@@ -81,7 +81,7 @@ export default (server: McpServer) => {
       try {
         sampleLines = (await axios.get(
           sampleUrl.toString(),
-          buildAxiosOptions(extra.requestInfo?.headers)
+          buildAxiosOptions(baseUrl)
         )).data.results
       } catch (err: any) {
         handleApiError(err)

src/mcp-servers/datasets/tools/get-field-values.ts

@@ -48,7 +48,7 @@ export default (server: McpServer) => {
       try {
         values = (await axios.get(
           fetchUrl.toString(),
-          buildAxiosOptions(extra.requestInfo?.headers)
+          buildAxiosOptions(baseUrl)
         )).data
       } catch (err: any) {
         handleApiError(err)

src/mcp-servers/datasets/tools/search-data.ts

@@ -71,7 +71,7 @@ export default (server: McpServer) => {
       try {
         response = (await axios.get(
           fetchUrlStr,
-          buildAxiosOptions(extra.requestInfo?.headers)
+          buildAxiosOptions(baseUrl)
         )).data
       } catch (err: any) {
         handleApiError(err)

src/mcp-servers/datasets/tools/search-datasets.ts

@@ -45,7 +45,7 @@ export default (server: McpServer) => {
       try {
         fetchedData = (await axios.get(
           fetchUrl.toString(),
-          buildAxiosOptions(extra.requestInfo?.headers)
+          buildAxiosOptions(baseUrl)
         )).data
       } catch (err: any) {
         handleApiError(err)