mcp-scraper 0.61.0 → 0.62.0

package/dist/bin/mcp-stdio-server.js

@@ -13,7 +13,7 @@ import {
   registerScheduledResultsMcpTools,
   registerSerpIntelligenceCaptureTools,
   resolveDeploymentProfile
-} from "../chunk-CIKVUFC5.js";
+} from "../chunk-Y2RIYBYR.js";
 import "../chunk-2K74LVV7.js";
 import "../chunk-SQMKUPA5.js";
 import "../chunk-57Y5QHI3.js";
@@ -22,7 +22,7 @@ import {
 } from "../chunk-IMBMDZUO.js";
 import {
   PACKAGE_VERSION
-} from "../chunk-E6QGM3ZA.js";
+} from "../chunk-RAPIYV2X.js";
 import "../chunk-2U2ZDDSI.js";
 import "../chunk-B7QNUVBT.js";
 import "../chunk-6OHW5XZN.js";

package/dist/chunk-RAPIYV2X.js

@@ -0,0 +1,7 @@
+// src/version.ts
+var PACKAGE_VERSION = "0.62.0";
+
+export {
+  PACKAGE_VERSION
+};
+//# sourceMappingURL=chunk-RAPIYV2X.js.map

package/dist/chunk-RAPIYV2X.js.map

@@ -0,0 +1 @@
+{"version":3,"sources":["../src/version.ts"],"sourcesContent":["export const PACKAGE_VERSION = '0.62.0'\n"],"mappings":";AAAO,IAAM,kBAAkB;","names":[]}

package/dist/{chunk-CIKVUFC5.js → chunk-Y2RIYBYR.js}

@@ -21,7 +21,7 @@ import {
 } from "./chunk-57Y5QHI3.js";
 import {
   PACKAGE_VERSION
-} from "./chunk-E6QGM3ZA.js";
+} from "./chunk-RAPIYV2X.js";
 import {
   MC_PER_CREDIT,
   PAA_BASE_CREDITS,
@@ -9577,7 +9577,7 @@ function registerPaaExtractorMcpTools(server, executor, options = {}) {
   }, async (input) => formatSearchSerp(await executor.searchSerp(input), input));
   server.registerTool("extract_url", {
     title: "Single URL Extract",
-    description: "Extract structured data from one public URL: content, schema, headings, metadata, screenshots, branding, featured image, or media assets. Wayback replay URLs automatically return the archived page copy without playback chrome. Use delivery:auto for bounded inline results with automatic artifact offload, delivery:artifact for a durable owner-scoped report, or delivery:memory to save the full page into hosted MCP Memory. preserveMedia is the preferred media-retention flag; depositToVault and downloadMedia remain temporary compatibility aliases.",
+    description: "Extract structured data from one public URL: content, schema, headings, metadata, screenshots, branding, featured image, or media assets. Tries a plain HTTP fetch first and only opens a stealth browser when that fetch is blocked, hits a bot check, or returns unusably thin content \u2014 most calls never need the browser, and a bot_check_unresolved error means the browser already tried and failed, not that a retry will help. Wayback replay URLs automatically return the archived page copy without playback chrome. Use delivery:auto for bounded inline results with automatic artifact offload, delivery:artifact for a durable owner-scoped report, or delivery:memory to save the full page into hosted MCP Memory. preserveMedia is the preferred media-retention flag; depositToVault and downloadMedia remain temporary compatibility aliases.",
     inputSchema: exposesLocalNetworkAccess ? ExtractUrlLocalInputSchema : ExtractUrlInputSchema,
     outputSchema: recordOutputSchema("extract_url", ExtractUrlOutputSchema),
     annotations: { ...liveWebToolAnnotations("Single URL Extract"), readOnlyHint: false }
@@ -9605,7 +9605,7 @@ function registerPaaExtractorMcpTools(server, executor, options = {}) {
   }, async (input) => formatMapWaybackSnapshots(await executor.mapWaybackSnapshots(input), input, ctx));
   server.registerTool("extract_site", {
     title: "Multi-Page Site Content Crawl",
-    description: `Crawl a public website and return page CONTENT (Markdown) across multiple pages. A Wayback replay URL produces one archived site snapshot. The optional wayback plan produces whole-site, single-page, or selected-page timelines across explicit months or a month range, all in one export with a capture matrix. Pass a new idempotencyKey for each intended crawl and reuse it only when retrying that call. Every MCP crawl starts a durable export; poll check_site_export for honest outcome counters and ${fileBehavior("the saved ZIP.", "the owner-scoped downloadable ZIP.")} Content only \u2014 for a technical SEO audit use audit_site instead.`,
+    description: `Crawl a public website and return page CONTENT (Markdown) across multiple pages. Each page is fetched with a plain HTTP request first; a browser retries only the pages that were blocked or hit a bot check, so a per-page bot_check_unresolved result means that specific page already failed the browser retry too. A Wayback replay URL produces one archived site snapshot. The optional wayback plan produces whole-site, single-page, or selected-page timelines across explicit months or a month range, all in one export with a capture matrix. Pass a new idempotencyKey for each intended crawl and reuse it only when retrying that call. Every MCP crawl starts a durable export; poll check_site_export for honest outcome counters and ${fileBehavior("the saved ZIP.", "the owner-scoped downloadable ZIP.")} Content only \u2014 for a technical SEO audit use audit_site instead.`,
     inputSchema: ExtractSiteInputSchema,
     outputSchema: recordOutputSchema("extract_site", ExtractSiteOutputSchema),
     annotations: { ...liveWebToolAnnotations("Multi-Page Site Content Crawl"), readOnlyHint: false }
@@ -16008,4 +16008,4 @@ export {
   ScheduledResultsMcpExecutor,
   registerScheduledResultsMcpTools
 };
-//# sourceMappingURL=chunk-CIKVUFC5.js.map
+//# sourceMappingURL=chunk-Y2RIYBYR.js.map