mcp-scraper 0.32.2 → 0.33.0

package/dist/{chunk-XEED325T.js → chunk-JURXY3DP.js}

@@ -19,7 +19,7 @@ import {
 } from "./chunk-V73MPRU6.js";
 import {
   PACKAGE_VERSION
-} from "./chunk-XLO5CZ6T.js";
+} from "./chunk-IASKABFL.js";
 import {
   MC_PER_CREDIT
 } from "./chunk-4767BA2O.js";
@@ -3481,7 +3481,7 @@ var HarvestPaaInputSchema = {
   gl: z.string().length(2).default("us").describe("Google country code inferred from location or user language."),
   hl: z.string().default("en").describe("Google interface/content language inferred from the user request."),
   device: z.enum(["desktop", "mobile"]).default("desktop").describe("SERP device context. Use mobile only for mobile rankings."),
-  proxyMode: z.enum(["location", "configured", "none"]).default(DEFAULT_PROXY_MODE).describe('Leave unset (clean egress). Do NOT set "location" just because a city was named \u2014 that comes from city-in-query wording. "location" forces residential geo-IP for rank-tracking fidelity, is frequently CAPTCHA-blocked, and accepts failures.'),
+  proxyMode: z.enum(["configured", "none"]).default(DEFAULT_PROXY_MODE).describe("Leave unset for clean egress (the default). Country/region localization comes from gl/hl plus the city or region in the query \u2014 not from a proxy."),
   proxyZip: z.string().regex(/^\d{5}$/).optional().describe('US ZIP for residential geo-IP targeting. Only meaningful with proxyMode "location".'),
   debug: z.boolean().default(false).describe("Include sanitized diagnostics for debugging localization, CAPTCHA, or proxy behavior.")
 };
@@ -3650,7 +3650,7 @@ var MapsSearchInputSchema = {
   hl: z.string().length(2).default("en").describe("Language inferred from user request."),
   maxResults: z.number().int().min(1).max(50).default(10).describe("Number of candidates to return. Default 10, maximum 50."),
   includeServices: z.boolean().default(false).describe("Open each returned business profile to include its configured services and areas served when available. Adds a page visit per business; does not collect review cards."),
-  proxyMode: z.enum(["location", "configured", "none"]).default(DEFAULT_MAPS_PROXY_MODE).describe("Leave unset for the default route (stealth browser on the managed ISP proxy, retried on a fresh session when Google blocks). Localization comes from the city in the query plus gl/hl. location forces an explicit residential proxy \u2014 not recommended for Google."),
+  proxyMode: z.enum(["configured", "none"]).default(DEFAULT_MAPS_PROXY_MODE).describe("Leave unset for the default route (stealth browser on the managed ISP proxy, retried on a fresh session when Google blocks). Country/region localization comes from the city or region in the query plus gl/hl \u2014 not from a proxy."),
   proxyZip: z.string().regex(/^\d{5}$/).optional().describe("Optional US ZIP override, only used when proxyMode is location."),
   debug: z.boolean().default(false).describe("Include sanitized browser/proxy diagnostics.")
 };
@@ -3665,7 +3665,7 @@ var DirectoryWorkflowInputSchema = {
   includeZipGroups: z.boolean().default(true).describe("Attach ZIP groups from a configured US ZIPS CSV when available (MCP_SCRAPER_USZIPS_CSV_PATH or usZipsCsvPath)."),
   usZipsCsvPath: z.string().optional().describe("Local/test-only path to a US ZIPS CSV (state_abbr, zipcode, county, city columns). Deployed APIs should use MCP_SCRAPER_USZIPS_CSV_PATH instead. For ZIP enrichment, set MCP_SCRAPER_USZIPS_CSV_PATH on the server, or pass this in local/test mode."),
   saveCsv: z.boolean().default(true).describe("Save a directory-ready CSV of results to the MCP Scraper output directory and return its path."),
-  proxyMode: z.enum(["location", "configured", "none"]).default(DEFAULT_MAPS_PROXY_MODE).describe("Proxy behavior per city search. Leave unset for the default route (stealth browser on the managed ISP proxy, retried fresh on a Google block). location forces an explicit residential proxy \u2014 not recommended for Google."),
+  proxyMode: z.enum(["configured", "none"]).default(DEFAULT_MAPS_PROXY_MODE).describe("Proxy behavior per city search. Leave unset for the default route (stealth browser on the managed ISP proxy, retried fresh on a Google block). Country/region localization comes from the city or region in the query plus gl/hl \u2014 not from a proxy."),
   proxyZip: z.string().regex(/^\d{5}$/).optional().describe("Optional ZIP override for proxy targeting; normally omitted."),
   debug: z.boolean().default(false).describe("Include sanitized browser/proxy diagnostics.")
 };
@@ -4505,7 +4505,7 @@ var SearchSerpInputSchema = {
   gl: z.string().length(2).default("us").describe("Google country code inferred from location or user language."),
   hl: z.string().default("en").describe("Google interface/content language inferred from user request."),
   device: z.enum(["desktop", "mobile"]).default("desktop").describe("SERP device context. Use mobile only for mobile rankings."),
-  proxyMode: z.enum(["location", "configured", "none"]).default(DEFAULT_PROXY_MODE).describe('Leave unset (clean egress). Do NOT set "location" just because a city was named \u2014 that comes from city-in-query wording. "location" forces residential geo-IP for rank-tracking fidelity, is frequently CAPTCHA-blocked, and accepts failures.'),
+  proxyMode: z.enum(["configured", "none"]).default(DEFAULT_PROXY_MODE).describe("Leave unset for clean egress (the default). Country/region localization comes from gl/hl plus the city or region in the query \u2014 not from a proxy."),
   proxyZip: z.string().regex(/^\d{5}$/).optional().describe('US ZIP for residential geo-IP targeting. Only meaningful with proxyMode "location".'),
   debug: z.boolean().default(false).describe("Include sanitized diagnostics for debugging localization, CAPTCHA, or proxy behavior."),
   pages: z.number().int().min(1).max(2).default(1).describe("Number of result pages to fetch (1\u20132).")
@@ -4516,7 +4516,7 @@ var CaptureSerpSnapshotInputSchema = {
   gl: z.string().length(2).default("us").describe("Google country code inferred from the requested market."),
   hl: z.string().default("en").describe("Google interface/content language inferred from the user request."),
   device: z.enum(["desktop", "mobile"]).default("desktop").describe("SERP device context. Use mobile only for mobile rankings/evidence."),
-  proxyMode: z.enum(["location", "configured", "none"]).default(DEFAULT_PROXY_MODE).describe('Leave unset (clean egress). Do NOT set "location" just because a city was named \u2014 that comes from city-in-query wording. "location" forces residential geo-IP, is frequently CAPTCHA-blocked, and accepts failures.'),
+  proxyMode: z.enum(["configured", "none"]).default(DEFAULT_PROXY_MODE).describe("Leave unset for clean egress (the default). Country/region localization comes from gl/hl plus the city or region in the query \u2014 not from a proxy."),
   proxyZip: z.string().regex(/^\d{5}$/).optional().describe('US ZIP for residential geo-IP targeting. Only meaningful with proxyMode "location".'),
   pages: z.number().int().min(1).max(2).default(1).describe("Google result pages to capture. Use 2 only for deeper ranking evidence."),
   debug: z.boolean().default(false).describe("Include sanitized browser/proxy/location diagnostics."),
@@ -4875,6 +4875,37 @@ var GmailSendMessageOutputSchema = {
   result: z.unknown().optional(),
   error: NullableString
 };
+var GmailSearchContactsInputSchema = {
+  connectionId: z.string().min(1).describe("A Gmail connectionId from list_service_connections. Read-only \u2014 does not require actionsEnabled."),
+  query: z.string().min(1).max(500).describe('Gmail search syntax, e.g. "from:brandnorth.com after:2026/03/22" or "brandnorth.com".'),
+  maxMessages: z.number().int().min(1).max(150).default(50).describe("Max messages to fetch and aggregate in this call. Paginate with pageToken for more; the response reports totalMatches and truncated so undercoverage is never silent."),
+  pageToken: z.string().optional().describe("Continuation token from a prior response to fetch the next page.")
+};
+var GmailSearchContactsOutputSchema = {
+  ok: z.boolean(),
+  totalMatches: z.number().optional().describe("Gmail's estimated total matches for the query, independent of how many were fetched this call."),
+  messagesFetched: z.number().optional(),
+  truncated: z.boolean().optional().describe("True when totalMatches exceeds messagesFetched \u2014 more results exist, use nextPageToken."),
+  nextPageToken: NullableString,
+  messages: z.array(z.object({
+    id: z.string(),
+    date: NullableString,
+    from: NullableString,
+    to: NullableString,
+    subject: NullableString,
+    snippet: NullableString
+  })).optional(),
+  contacts: z.array(z.object({
+    email: z.string(),
+    name: NullableString,
+    domain: NullableString,
+    messageCount: z.number(),
+    firstSeen: NullableString,
+    lastSeen: NullableString,
+    sampleSubjects: z.array(z.string())
+  })).optional().describe("Messages deduped by sender email address, newest-to-oldest within this fetch only \u2014 not the full query result set when truncated."),
+  error: NullableString
+};
 var GoogleCalendarCreateEventInputSchema = {
   connectionId: z.string().min(1).describe("A Google Calendar connectionId from list_service_connections, with actionsEnabled true."),
   calendarId: z.string().min(1).default("primary").describe('Calendar to create the event in. Default "primary".'),
@@ -5613,6 +5644,13 @@ function registerPaaExtractorMcpTools(server, executor, options = {}) {
     outputSchema: recordOutputSchema("gmail_send_message", GmailSendMessageOutputSchema),
     annotations: { title: "Send Gmail Message", readOnlyHint: false, destructiveHint: false, idempotentHint: false, openWorldHint: true }
   }, async (input) => executor.gmailSendMessage(input));
+  server.registerTool("gmail_search_contacts", {
+    title: "Search Gmail Contacts",
+    description: 'Search Gmail with standard Gmail query syntax (e.g. "from:acme.com after:2026/03/22") and get back deduplicated sender contacts (email, name, domain, message count, first/last seen, sample subjects) instead of raw messages. Read-only \u2014 works on any connected Gmail connection from list_service_connections, no actionsEnabled required. Use this instead of looping list-messages/get-message yourself: those return bare message IDs and full raw MIME per message, which does not scale past a handful of messages. Reports totalMatches and truncated so incomplete coverage from a large result set is never silent \u2014 pass the returned nextPageToken to continue.',
+    inputSchema: GmailSearchContactsInputSchema,
+    outputSchema: recordOutputSchema("gmail_search_contacts", GmailSearchContactsOutputSchema),
+    annotations: { title: "Search Gmail Contacts", readOnlyHint: true, destructiveHint: false, idempotentHint: true, openWorldHint: true }
+  }, async (input) => executor.gmailSearchContacts(input));
   server.registerTool("google_calendar_create_event", {
     title: "Create Calendar Event",
     description: "Create a complete event on a connected, action-enabled Google Calendar connection. Always preserve the supplied purpose in description, include the Zoom join link when available, and include every explicitly named invitee in attendees. Do not create a bare meeting event. Requires a connectionId from list_service_connections with actionsEnabled true.",
@@ -5997,6 +6035,9 @@ var HttpMcpToolExecutor = class {
   gmailSendMessage(input) {
     return this.call("/schedule-connections/actions/gmail/send-message", input);
   }
+  gmailSearchContacts(input) {
+    return this.call("/schedule-connections/actions/gmail/search-contacts", input);
+  }
   googleCalendarCreateEvent(input) {
     return this.call("/schedule-connections/actions/google-calendar/create-event", input);
   }
@@ -10110,4 +10151,4 @@ export {
   MEMORY_TOOL_SCHEMAS,
   registerMemoryMcpTools
 };
-//# sourceMappingURL=chunk-XEED325T.js.map
+//# sourceMappingURL=chunk-JURXY3DP.js.map