mcp-scraper 0.66.3 → 0.66.5

package/dist/bin/mcp-scraper-install.cjs

@@ -3,7 +3,7 @@
 
 // release-message.json
 var release_message_default = {
-  message: "Strict MCP clients now discover all 259 tools through a compact runtime catalog, while generated SDKs retain complete output types."
+  message: "Local Sourcebook submissions now clearly distinguish processing from publication and return the exact live listing URL when complete."
 };
 
 // src/install-terminal.ts
@@ -112,7 +112,7 @@ function renderInstallTerminal(options) {
 }
 
 // src/version.ts
-var PACKAGE_VERSION = "0.66.3";
+var PACKAGE_VERSION = "0.66.5";
 
 // bin/mcp-scraper-install.ts
 var noColor = process.argv.includes("--no-color") || process.env.NO_COLOR !== void 0 || process.env.FORCE_COLOR === "0" || !process.stdout.isTTY;

package/dist/bin/mcp-scraper-install.js

@@ -1,10 +1,10 @@
 #!/usr/bin/env node
 import {
   renderInstallTerminal
-} from "../chunk-2XVDLF6H.js";
+} from "../chunk-TCOP5BT7.js";
 import {
   PACKAGE_VERSION
-} from "../chunk-JBFLOJGM.js";
+} from "../chunk-RSFMR7O7.js";
 
 // bin/mcp-scraper-install.ts
 var noColor = process.argv.includes("--no-color") || process.env.NO_COLOR !== void 0 || process.env.FORCE_COLOR === "0" || !process.stdout.isTTY;

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

@@ -1232,7 +1232,7 @@ render();
 }
 
 // src/version.ts
-var PACKAGE_VERSION = "0.66.3";
+var PACKAGE_VERSION = "0.66.5";
 
 // src/mcp/input-field-descriptions.ts
 var import_zod = require("zod");
@@ -8378,11 +8378,16 @@ Multi-step orchestrations \u2014 prefer these over hand-chaining primitives when
 - A new capture costs 10 Credits total and includes broad website, exact Maps place, maximum-accessible review,
   service-area, staff, and genuine-media acquisition. When compilation succeeds, that exact system-authored evidence revision
   publishes automatically. Never supply owner-authored services, reviews, images, or other public claims.
+- An accepted capture is still processing, not complete. While its returned \`statusCheck.required\` is true,
+  wait the suggested interval and call **local_sourcebook_submission_status** again before giving the user a
+  final completion answer. When it publishes, say that the listing is live and include the exact clickable
+  \`publicUrls.profileUrl\`; when it fails, report the returned failure message and cause.
 - Read progress and the compiled record with **local_sourcebook_submission_status**. Use
   **local_sourcebook_refresh** only when the owner intentionally wants to pay 2 Credits for a new acquisition pass.
   Contract, tag, preparation, validation, and status tools are free; idempotent retries are not charged twice.
-- Capture and status responses return \`publicUrls.profileUrl\` and \`publicUrls.reviewsUrl\`. Always give those
-  exact links to the user; do not report a listing as live without including its clickable profile URL.
+- Capture and status responses return a plain-language \`message\`, a \`statusCheck\`, and exact public URLs. Do not
+  describe a queued or enriching listing as complete, and do not report a listing as live without including its
+  clickable \`publicUrls.profileUrl\`.
 - The last published revision remains public during refresh. Administrators can reject or unpublish an
   exceptional record, but routine subscriber publication does not require administrator review.
 
@@ -8408,10 +8413,10 @@ Multi-step orchestrations \u2014 prefer these over hand-chaining primitives when
   artifact, call \`report_artifact_read\` with the returned artifactId and follow nextOffset until null. This
   works through the authenticated MCP connection even when the client cannot open the signed download URL;
   do not try curl or web_fetch. Use \`archive_read\` for ZIP archives.
-- Before using a connected account, call \`list_service_connections\` and match the intended provider-side
-  identity from \`providerAccountEmail\` or \`providerAccountName\`, not the MCP Scraper login. If
-  \`providerIdentityStatus\` is \`unavailable\`, ask the person to refresh that connection before assuming
-  which external account it represents.
+- Before using a connected account, call \`list_service_connections\`. Match an explicit \`userLabel\` first,
+  then verified provider-side \`providerAccountEmail\`, \`providerAccountName\`, or \`providerAccountId\`; never
+  treat the MCP Scraper login as external-provider identity. If identity is pending or unavailable and no
+  distinct user label exists, ask the person which connection they mean instead of assuming.
 - For Google Search Console, use \`export_connected_service_data\` with
   \`dataset:"search_console_performance"\` when the person asks for a fresh JSONL download. When they
   ask to schedule ingestion or repeatedly filter historical rows, create a \`connection_sync\` schedule,
@@ -12064,7 +12069,12 @@ var ListServiceConnectionsOutputSchema = {
     connectionId: import_zod8.z.string(),
     providerConfigKey: import_zod8.z.string(),
     provider: import_zod8.z.string().nullable().optional(),
-    label: import_zod8.z.string(),
+    label: import_zod8.z.string().describe("Best available human-facing account label: a userLabel when set, otherwise verified provider identity, then a stable fallback."),
+    userLabel: import_zod8.z.string().nullable().optional().describe("Optional tenant-owned name set in the dashboard to distinguish this connection. This is not asserted provider identity."),
+    providerAccountId: import_zod8.z.string().nullable().optional().describe("Stable provider-side account, user, workspace, channel, or tenant identifier when the provider exposes one."),
+    providerAccountEmail: import_zod8.z.string().email().nullable().optional().describe("Verified provider-side email when the provider exposes one. This is never inferred from the MCP Scraper login."),
+    providerAccountName: import_zod8.z.string().nullable().optional().describe("Provider-side account, user, workspace, channel, or tenant name when the provider exposes one."),
+    providerIdentityStatus: import_zod8.z.enum(["pending", "verified", "unavailable"]).optional().describe("Whether provider identity discovery is pending, returned at least one verified provider identifier, or completed without one."),
     status: import_zod8.z.string(),
     lifecycleStatus: import_zod8.z.enum(["pending", "connected", "needs_reauth", "disconnecting", "disconnected"]).optional().describe("Credential lifecycle. This is separate from current provider availability."),
     operationalStatus: import_zod8.z.enum(["unknown", "available", "degraded", "unavailable"]).optional().describe("Last observed provider transport availability. Unavailable does not imply reconnect is required."),
@@ -14226,14 +14236,14 @@ function registerPaaExtractorMcpTools(server, executor, options = {}) {
   }, async (input) => executor.validateLocalSourcebookWrite(input));
   server.registerTool("local-sourcebook-capture", {
     title: "Capture Governed Local Sourcebook Listing",
-    description: "Strict owner-scoped LocalBusiness write path after list, contract, prepare, and validate. Creating a listing costs 10 Credits total, including website, exact-place, review, service-area, staff, genuine-media, and structured-data compilation; preparation and validation are free, and idempotent retries are not charged twice. Capture registers canonical tags and the accepted Schema.org LocalBusiness subtype, returns the exact LocalSourcebook.com profile and reviews URLs, and queues acquisition. MCP Scraper generates the final JSON-LD from source-backed evidence. A successful system-compiled evidence revision publishes automatically; owner-authored public claims and arbitrary JSON-LD are not accepted.",
+    description: "Strict owner-scoped LocalBusiness write path after list, contract, prepare, and validate. Creating a listing costs 10 Credits total, including website, exact-place, review, service-area, staff, genuine-media, and structured-data compilation; preparation and validation are free, and idempotent retries are not charged twice. Capture registers canonical tags and the accepted Schema.org LocalBusiness subtype, then queues acquisition; an accepted capture is not yet live. Poll local_sourcebook_submission_status using statusCheck until it is terminal. When published, tell the user the returned message and exact publicUrls.profileUrl. MCP Scraper generates the final JSON-LD from source-backed evidence and publishes a successful system-compiled revision automatically; owner-authored public claims and arbitrary JSON-LD are not accepted.",
     inputSchema: LocalSourcebookCaptureInputSchema,
     outputSchema: recordOutputSchema("local-sourcebook-capture", LocalSourcebookOutputSchema),
     annotations: liveWebToolAnnotations("Capture Governed Local Sourcebook Listing")
   }, async (input) => executor.localSourcebookCapture(input));
   server.registerTool("local_sourcebook_submission_status", {
     title: "Local Sourcebook Submission Status",
-    description: "Read the caller\u2019s listing draft, enrichment coverage, revision number, publication state, and live LocalSourcebook.com profile and reviews URLs. When status is failed, acquisitionError says why \u2014 read it before retrying, since a refresh repeats the same acquisition. Coverage counters are written only on success, so zeroed counters on a failed listing do not mean the crawl never ran.",
+    description: "Read the caller\u2019s listing draft, enrichment coverage, revision number, publication state, and LocalSourcebook.com profile and reviews URLs. While statusCheck.required is true, wait the returned interval and check again before giving the user a final completion answer. When published, present the returned message and exact clickable publicUrls.profileUrl. When failed, present the failure message and read acquisitionError before retrying, since a refresh repeats the same acquisition. Coverage counters are written only on success, so zeroed counters on a failed listing do not mean the crawl never ran.",
     inputSchema: LocalSourcebookSubmissionStatusInputSchema,
     outputSchema: recordOutputSchema("local_sourcebook_submission_status", LocalSourcebookOutputSchema),
     annotations: localPlanningToolAnnotations("Local Sourcebook Submission Status")
@@ -14378,7 +14388,7 @@ function registerPaaExtractorMcpTools(server, executor, options = {}) {
   }, async (input) => formatCreditsInfo(await executor.creditsInfo(input), input));
   server.registerTool("list_service_connections", {
     title: "List Connected Services",
-    description: "List every service connection this MCP Scraper account has authorized, including Resend, GitHub, Google Analytics, Google Search Console, YouTube, Facebook Pages, LinkedIn, X, Meta Marketing, Slack, Gmail, Calendar, Google Drive, Zoom, Xero, and others. Returns the tenant-scoped connectionId, credential transport, exact live readTools and gated actionTools, permission-aware toolCapabilities with missing OAuth-grant or provider-app-feature blockers, permanently blocked administrative tools, and schema-discovery metadata. Get a connectionId and exact tool name here before calling describe_service_connection_tool, read_service_connection, or call_service_connection_action. Managed OAuth and official remote MCP connections use the same provider-neutral bridges; mutations still require the account action switch and an exact allowed action. A scheduled Search Console connection_sync creates a typed tenant-owned performance table; after it runs, use the returned tableName with table-describe and table-query instead of repeatedly calling Google for historical filtering.",
+    description: "List every service connection this MCP Scraper account has authorized, including Resend, GitHub, Google Analytics, Google Search Console, YouTube, Facebook Pages, LinkedIn, X, Meta Marketing, Slack, Gmail, Calendar, Google Drive, Zoom, Xero, and others. Returns the tenant-scoped connectionId; a tenant-owned userLabel; verified provider-side account ID, email, or name when available; credential transport; exact live readTools and gated actionTools; permission-aware toolCapabilities with missing OAuth-grant or provider-app-feature blockers; permanently blocked administrative tools; and schema-discovery metadata. Get a connectionId and exact tool name here before calling describe_service_connection_tool, read_service_connection, or call_service_connection_action. Managed OAuth and official remote MCP connections use the same provider-neutral bridges; mutations still require the account action switch and an exact allowed action. A scheduled Search Console connection_sync creates a typed tenant-owned performance table; after it runs, use the returned tableName with table-describe and table-query instead of repeatedly calling Google for historical filtering.",
     inputSchema: ListServiceConnectionsInputSchema,
     outputSchema: recordOutputSchema("list_service_connections", ListServiceConnectionsOutputSchema),
     annotations: { title: "List Connected Services", readOnlyHint: true, destructiveHint: false, idempotentHint: true, openWorldHint: false }
@@ -18655,7 +18665,7 @@ var HostedMemoryMcpToolExecutor = class {
 
 // release-message.json
 var release_message_default = {
-  message: "Strict MCP clients now discover all 259 tools through a compact runtime catalog, while generated SDKs retain complete output types."
+  message: "Local Sourcebook submissions now clearly distinguish processing from publication and return the exact live listing URL when complete."
 };
 
 // src/install-terminal.ts

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

@@ -13,18 +13,18 @@ import {
   registerScheduledResultsMcpTools,
   registerSerpIntelligenceCaptureTools,
   resolveDeploymentProfile
-} from "../chunk-LNLMOBP6.js";
+} from "../chunk-OJFWURGD.js";
 import "../chunk-PGJQDMC2.js";
 import "../chunk-3EYALCYK.js";
 import "../chunk-3TKKJQB4.js";
 import {
   renderInstallTerminal
-} from "../chunk-2XVDLF6H.js";
+} from "../chunk-TCOP5BT7.js";
 import "../chunk-VXLU74YZ.js";
 import "../chunk-GGZEC22A.js";
 import {
   PACKAGE_VERSION
-} from "../chunk-JBFLOJGM.js";
+} from "../chunk-RSFMR7O7.js";
 import "../chunk-DNM65UCK.js";
 import "../chunk-AYJHI2FK.js";
 import "../chunk-55B3T5L2.js";

package/dist/{chunk-LNLMOBP6.js → chunk-OJFWURGD.js}

@@ -25,7 +25,7 @@ import {
 } from "./chunk-VXLU74YZ.js";
 import {
   PACKAGE_VERSION
-} from "./chunk-JBFLOJGM.js";
+} from "./chunk-RSFMR7O7.js";
 import {
   createPrivateArtifact,
   privateArtifactOwnerId,
@@ -550,11 +550,16 @@ Multi-step orchestrations \u2014 prefer these over hand-chaining primitives when
 - A new capture costs 10 Credits total and includes broad website, exact Maps place, maximum-accessible review,
   service-area, staff, and genuine-media acquisition. When compilation succeeds, that exact system-authored evidence revision
   publishes automatically. Never supply owner-authored services, reviews, images, or other public claims.
+- An accepted capture is still processing, not complete. While its returned \`statusCheck.required\` is true,
+  wait the suggested interval and call **local_sourcebook_submission_status** again before giving the user a
+  final completion answer. When it publishes, say that the listing is live and include the exact clickable
+  \`publicUrls.profileUrl\`; when it fails, report the returned failure message and cause.
 - Read progress and the compiled record with **local_sourcebook_submission_status**. Use
   **local_sourcebook_refresh** only when the owner intentionally wants to pay 2 Credits for a new acquisition pass.
   Contract, tag, preparation, validation, and status tools are free; idempotent retries are not charged twice.
-- Capture and status responses return \`publicUrls.profileUrl\` and \`publicUrls.reviewsUrl\`. Always give those
-  exact links to the user; do not report a listing as live without including its clickable profile URL.
+- Capture and status responses return a plain-language \`message\`, a \`statusCheck\`, and exact public URLs. Do not
+  describe a queued or enriching listing as complete, and do not report a listing as live without including its
+  clickable \`publicUrls.profileUrl\`.
 - The last published revision remains public during refresh. Administrators can reject or unpublish an
   exceptional record, but routine subscriber publication does not require administrator review.
 
@@ -580,10 +585,10 @@ Multi-step orchestrations \u2014 prefer these over hand-chaining primitives when
   artifact, call \`report_artifact_read\` with the returned artifactId and follow nextOffset until null. This
   works through the authenticated MCP connection even when the client cannot open the signed download URL;
   do not try curl or web_fetch. Use \`archive_read\` for ZIP archives.
-- Before using a connected account, call \`list_service_connections\` and match the intended provider-side
-  identity from \`providerAccountEmail\` or \`providerAccountName\`, not the MCP Scraper login. If
-  \`providerIdentityStatus\` is \`unavailable\`, ask the person to refresh that connection before assuming
-  which external account it represents.
+- Before using a connected account, call \`list_service_connections\`. Match an explicit \`userLabel\` first,
+  then verified provider-side \`providerAccountEmail\`, \`providerAccountName\`, or \`providerAccountId\`; never
+  treat the MCP Scraper login as external-provider identity. If identity is pending or unavailable and no
+  distinct user label exists, ask the person which connection they mean instead of assuming.
 - For Google Search Console, use \`export_connected_service_data\` with
   \`dataset:"search_console_performance"\` when the person asks for a fresh JSONL download. When they
   ask to schedule ingestion or repeatedly filter historical rows, create a \`connection_sync\` schedule,
@@ -9442,7 +9447,12 @@ var ListServiceConnectionsOutputSchema = {
     connectionId: z6.string(),
     providerConfigKey: z6.string(),
     provider: z6.string().nullable().optional(),
-    label: z6.string(),
+    label: z6.string().describe("Best available human-facing account label: a userLabel when set, otherwise verified provider identity, then a stable fallback."),
+    userLabel: z6.string().nullable().optional().describe("Optional tenant-owned name set in the dashboard to distinguish this connection. This is not asserted provider identity."),
+    providerAccountId: z6.string().nullable().optional().describe("Stable provider-side account, user, workspace, channel, or tenant identifier when the provider exposes one."),
+    providerAccountEmail: z6.string().email().nullable().optional().describe("Verified provider-side email when the provider exposes one. This is never inferred from the MCP Scraper login."),
+    providerAccountName: z6.string().nullable().optional().describe("Provider-side account, user, workspace, channel, or tenant name when the provider exposes one."),
+    providerIdentityStatus: z6.enum(["pending", "verified", "unavailable"]).optional().describe("Whether provider identity discovery is pending, returned at least one verified provider identifier, or completed without one."),
     status: z6.string(),
     lifecycleStatus: z6.enum(["pending", "connected", "needs_reauth", "disconnecting", "disconnected"]).optional().describe("Credential lifecycle. This is separate from current provider availability."),
     operationalStatus: z6.enum(["unknown", "available", "degraded", "unavailable"]).optional().describe("Last observed provider transport availability. Unavailable does not imply reconnect is required."),
@@ -11735,14 +11745,14 @@ function registerPaaExtractorMcpTools(server, executor, options = {}) {
   }, async (input) => executor.validateLocalSourcebookWrite(input));
   server.registerTool("local-sourcebook-capture", {
     title: "Capture Governed Local Sourcebook Listing",
-    description: "Strict owner-scoped LocalBusiness write path after list, contract, prepare, and validate. Creating a listing costs 10 Credits total, including website, exact-place, review, service-area, staff, genuine-media, and structured-data compilation; preparation and validation are free, and idempotent retries are not charged twice. Capture registers canonical tags and the accepted Schema.org LocalBusiness subtype, returns the exact LocalSourcebook.com profile and reviews URLs, and queues acquisition. MCP Scraper generates the final JSON-LD from source-backed evidence. A successful system-compiled evidence revision publishes automatically; owner-authored public claims and arbitrary JSON-LD are not accepted.",
+    description: "Strict owner-scoped LocalBusiness write path after list, contract, prepare, and validate. Creating a listing costs 10 Credits total, including website, exact-place, review, service-area, staff, genuine-media, and structured-data compilation; preparation and validation are free, and idempotent retries are not charged twice. Capture registers canonical tags and the accepted Schema.org LocalBusiness subtype, then queues acquisition; an accepted capture is not yet live. Poll local_sourcebook_submission_status using statusCheck until it is terminal. When published, tell the user the returned message and exact publicUrls.profileUrl. MCP Scraper generates the final JSON-LD from source-backed evidence and publishes a successful system-compiled revision automatically; owner-authored public claims and arbitrary JSON-LD are not accepted.",
     inputSchema: LocalSourcebookCaptureInputSchema,
     outputSchema: recordOutputSchema("local-sourcebook-capture", LocalSourcebookOutputSchema),
     annotations: liveWebToolAnnotations("Capture Governed Local Sourcebook Listing")
   }, async (input) => executor.localSourcebookCapture(input));
   server.registerTool("local_sourcebook_submission_status", {
     title: "Local Sourcebook Submission Status",
-    description: "Read the caller\u2019s listing draft, enrichment coverage, revision number, publication state, and live LocalSourcebook.com profile and reviews URLs. When status is failed, acquisitionError says why \u2014 read it before retrying, since a refresh repeats the same acquisition. Coverage counters are written only on success, so zeroed counters on a failed listing do not mean the crawl never ran.",
+    description: "Read the caller\u2019s listing draft, enrichment coverage, revision number, publication state, and LocalSourcebook.com profile and reviews URLs. While statusCheck.required is true, wait the returned interval and check again before giving the user a final completion answer. When published, present the returned message and exact clickable publicUrls.profileUrl. When failed, present the failure message and read acquisitionError before retrying, since a refresh repeats the same acquisition. Coverage counters are written only on success, so zeroed counters on a failed listing do not mean the crawl never ran.",
     inputSchema: LocalSourcebookSubmissionStatusInputSchema,
     outputSchema: recordOutputSchema("local_sourcebook_submission_status", LocalSourcebookOutputSchema),
     annotations: localPlanningToolAnnotations("Local Sourcebook Submission Status")
@@ -11887,7 +11897,7 @@ function registerPaaExtractorMcpTools(server, executor, options = {}) {
   }, async (input) => formatCreditsInfo(await executor.creditsInfo(input), input));
   server.registerTool("list_service_connections", {
     title: "List Connected Services",
-    description: "List every service connection this MCP Scraper account has authorized, including Resend, GitHub, Google Analytics, Google Search Console, YouTube, Facebook Pages, LinkedIn, X, Meta Marketing, Slack, Gmail, Calendar, Google Drive, Zoom, Xero, and others. Returns the tenant-scoped connectionId, credential transport, exact live readTools and gated actionTools, permission-aware toolCapabilities with missing OAuth-grant or provider-app-feature blockers, permanently blocked administrative tools, and schema-discovery metadata. Get a connectionId and exact tool name here before calling describe_service_connection_tool, read_service_connection, or call_service_connection_action. Managed OAuth and official remote MCP connections use the same provider-neutral bridges; mutations still require the account action switch and an exact allowed action. A scheduled Search Console connection_sync creates a typed tenant-owned performance table; after it runs, use the returned tableName with table-describe and table-query instead of repeatedly calling Google for historical filtering.",
+    description: "List every service connection this MCP Scraper account has authorized, including Resend, GitHub, Google Analytics, Google Search Console, YouTube, Facebook Pages, LinkedIn, X, Meta Marketing, Slack, Gmail, Calendar, Google Drive, Zoom, Xero, and others. Returns the tenant-scoped connectionId; a tenant-owned userLabel; verified provider-side account ID, email, or name when available; credential transport; exact live readTools and gated actionTools; permission-aware toolCapabilities with missing OAuth-grant or provider-app-feature blockers; permanently blocked administrative tools; and schema-discovery metadata. Get a connectionId and exact tool name here before calling describe_service_connection_tool, read_service_connection, or call_service_connection_action. Managed OAuth and official remote MCP connections use the same provider-neutral bridges; mutations still require the account action switch and an exact allowed action. A scheduled Search Console connection_sync creates a typed tenant-owned performance table; after it runs, use the returned tableName with table-describe and table-query instead of repeatedly calling Google for historical filtering.",
     inputSchema: ListServiceConnectionsInputSchema,
     outputSchema: recordOutputSchema("list_service_connections", ListServiceConnectionsOutputSchema),
     annotations: { title: "List Connected Services", readOnlyHint: true, destructiveHint: false, idempotentHint: true, openWorldHint: false }

package/dist/{chunk-JBFLOJGM.js → chunk-RSFMR7O7.js}

@@ -1,5 +1,5 @@
 // src/version.ts
-var PACKAGE_VERSION = "0.66.3";
+var PACKAGE_VERSION = "0.66.5";
 
 export {
   PACKAGE_VERSION

package/dist/{chunk-2XVDLF6H.js → chunk-TCOP5BT7.js}

@@ -1,6 +1,6 @@
 // release-message.json
 var release_message_default = {
-  message: "Strict MCP clients now discover all 259 tools through a compact runtime catalog, while generated SDKs retain complete output types."
+  message: "Local Sourcebook submissions now clearly distinguish processing from publication and return the exact live listing URL when complete."
 };
 
 // src/install-terminal.ts