mcp-scraper 0.34.0 → 0.34.1

package/README.md

@@ -88,7 +88,7 @@ Build the branded one-click bundle:
 npm run build:mcpb
 ```
 
-The generated bundle is written to `build/mcpb/mcp-scraper-<version>.mcpb` and copied to `public/downloads/` for the hosted download. The current public bundle is `https://mcpscraper.dev/downloads/mcp-scraper.mcpb` (`0.33.7`, SHA-256 `0aee7cef4fd40fe3eb1db14788937efc3bd559557e68f608e9dda11e0fa1cb90`). Install it by opening or dragging it into Claude Desktop. Claude displays the `MCP Scraper` install card, icon, and API-key configuration field from the bundle manifest.
+The generated bundle is written to `build/mcpb/mcp-scraper-<version>.mcpb` and copied to `public/downloads/` for the hosted download. The current public bundle is `https://mcpscraper.dev/downloads/mcp-scraper.mcpb` (`0.34.1`, SHA-256 `b217a7273ae7ee406003619de7f4bf95b4c8ad241c7f723244187c1bb557d947`). Install it by opening or dragging it into Claude Desktop. Claude displays the `MCP Scraper` install card, icon, and API-key configuration field from the bundle manifest.
 
 The MCPB install exposes every tool — web-intelligence plus all `browser_*` tools — through the one `mcp-scraper` server.
 
@@ -213,7 +213,7 @@ Google Search Console exposes eight bounded reads and eight gated property and s
 - `browser_replay_download` — download and save the replay MP4 under `MCP_SCRAPER_OUTPUT_DIR/browser-replays`.
 - `browser_replay_mark` — while recording, locate a DOM target and return a replay-timed annotation object.
 - `browser_replay_annotate` — download a replay MP4, render timed boxes, circles, underlines, arrows, and labels using annotation objects from `browser_replay_mark` or exact bounds from `browser_locate`, and save a new annotated MP4 under `MCP_SCRAPER_OUTPUT_DIR/browser-replays`.
-- `browser_capture_fanout` — capture ChatGPT/Claude AI-search fan-out from an open logged-in hosted session. Use `export=true` for durable artifacts. It writes full `fanout.json`, query/source/citation/domain/snippet CSVs, TSV, and `report.html` under `MCP_SCRAPER_OUTPUT_DIR/fanout`, returning paths relative to `MCP_SCRAPER_OUTPUT_DIR` so users can share portable paths.
+- `browser_capture_fanout` — capture ChatGPT/Claude AI-search fan-out from an open logged-in hosted session. Every client receives the complete structured capture inline. Installed stdio/MCPB clients can use `export=true` for durable `fanout.json`, query/source/citation/domain/snippet CSVs, TSV, and `report.html` under `MCP_SCRAPER_OUTPUT_DIR/fanout`; hosted OAuth clients receive `exports: null` and use the inline data.
 - `browser_close`
 - `browser_list_sessions`
 
@@ -227,7 +227,7 @@ The `mcp-scraper` server (and the MCPB bundle, which runs it) exposes both secti
 
 All MCP tools expose output schemas and return `structuredContent` with the IDs, URLs, CSV paths, transcripts, browser session handles, replay paths, artifacts, recipe fields, or blueprint fields needed by the next step. Browser Agent tools keep a JSON text block for older clients, but structured data is the primary contract. All tools carry MCP annotations; file-writing tools such as replay downloads and annotations state their filesystem side effects.
 
-The canonical tool inventory is generated at `docs/mcp-tool-manifest.generated.json`. Both the `mcp-scraper` stdio server and the hosted endpoint at `https://mcpscraper.dev/mcp` expose the same 166 tools: 77 scraper, browser, workflow, billing, and connected-service tools plus 89 durable-memory tools. Release verification compares the exact local and remote tool-name sets, not only the count.
+The canonical tool inventory is generated at `docs/mcp-tool-manifest.generated.json`. Both the `mcp-scraper` stdio server and the hosted endpoint at `https://mcpscraper.dev/mcp` expose the same 167 tools: 78 scraper, browser, workflow, billing, and connected-service tools plus 89 durable-memory tools. Release verification compares the exact local and remote tool-name sets, not only the count.
 
 For contract parity, stdio and MCPB memory calls invoke the matching public tool on the hosted MCP Scraper `/mcp` endpoint. The hosted aggregate runtime owns MCP Scraper-specific billing, scheduling, credential, and in-process cutover policy; its internal `/memory/mcp-call` bridge is a fallback to the standalone memory service, not the public stdio execution path. Direct `mcp-memory` OAuth and stdio clients continue to use `memory.mcpscraper.dev` and must be verified as a separate dependent release surface.
 
@@ -245,7 +245,7 @@ The `mcp-scraper` NPX stdio server also exposes saved reports as MCP resources:
 - `BROWSER_AGENT_PROFILE_NAME` is optional and sets the default saved hosted browser profile for `mcp-scraper` stdio sessions. Aliases: `BROWSER_SERVICE_PROFILE_NAME`, `KERNEL_BROWSER_PROFILE_NAME`, `KERNEL_PROFILE_NAME`.
 - `BROWSER_AGENT_PROFILE_SAVE_CHANGES=true` is optional hosted setup behavior. It persists cookies and storage back to the named profile when `browser_close` deletes the hosted browser session. Aliases: `BROWSER_SERVICE_PROFILE_SAVE_CHANGES`, `KERNEL_BROWSER_PROFILE_SAVE_CHANGES`, `KERNEL_PROFILE_SAVE_CHANGES`.
 
-Every web intelligence tool call made through `mcp-scraper` saves a full Markdown report to disk by default and returns the file path in the MCP response. The hosted `/mcp` endpoint returns reports inline only and never writes files. Browser replay downloads are saved by `browser_replay_download` under `MCP_SCRAPER_OUTPUT_DIR/browser-replays`. AI fan-out exports are saved by `browser_capture_fanout` with `export=true` under `MCP_SCRAPER_OUTPUT_DIR/fanout`, and return relative paths.
+Every web intelligence tool call made through `mcp-scraper` saves a full Markdown report to disk by default and returns the file path in the MCP response. The hosted `/mcp` endpoint returns reports inline only and never writes files. Browser replay downloads are saved by `browser_replay_download` under `MCP_SCRAPER_OUTPUT_DIR/browser-replays`. AI fan-out captures are always returned inline; only installed stdio/MCPB clients write optional `export=true` files under `MCP_SCRAPER_OUTPUT_DIR/fanout`, returning relative paths. Hosted clients always receive `exports: null`.
 
 ## Updating Existing Installs
 

package/dist/bin/api-server.cjs

@@ -9166,7 +9166,11 @@ function insufficientBalanceResponse(balanceMc, requiredMc) {
     topup_url: topupUrl
   };
 }
-var MC_COSTS, MC_PER_BROWSER_MS, BROWSER_OPEN_MIN_BALANCE_MC, MC_PER_CREDIT, BILLING_MC_PER_USD_NUMERATOR, BILLING_MC_PER_USD_DENOMINATOR, BILLING_BASIS_POINTS, USD_NANOS, SCHEDULED_RUN_BASE_CREDITS, SCHEDULED_RUN_BASE_MC, SCHEDULED_AGENT_LLM_MARKUP_BPS, SCHEDULED_AGENT_LLM_MARKUP, SCHEDULED_RUN_DEFAULT_MAX_CREDITS, CONNECTED_USAGE_RATE_POLICY_VERSION, CONNECTED_ACTIVE_CONNECTION_MONTHLY_CREDITS, CONNECTED_ACTIVE_CONNECTION_MONTHLY_MC, CONNECTED_FUNCTION_RUN_CREDITS, CONNECTED_PROXY_REQUEST_CREDITS, CONNECTED_COMPUTE_CREDITS_PER_SECOND, CONNECTED_FUNCTION_RUN_MC, CONNECTED_PROXY_REQUEST_MC, CONNECTED_COMPUTE_MC_PER_SECOND, CONNECTED_USAGE_RATE_POLICY, SCHEDULED_ACTIONS_RATE_POLICY, CREDIT_COST_CATALOG, CONCURRENCY_PRICE_ID, SUBSCRIPTION_TIERS, SUBSCRIPTION_TIER_BY_KEY, PRIVATE_OFFERS, CONCURRENCY_SLOT_PRICE_USD, CONCURRENCY_SLOT_PRICE_CURRENCY, CONCURRENCY_SLOT_PRICE_INTERVAL, CONCURRENCY_SLOT_PRICE_LABEL, CONCURRENCY_SLOT_TERMINAL_COMMAND, MEMORY_FREE_QUOTA_BYTES, MEMORY_PLANS, MEMORY_PLAN_QUOTA, MEMORY_MARGIN_MULTIPLE, MEMORY_PRO_COST_BUDGET_USD, LedgerOperation, MEMORY_AI_MARGIN_MULTIPLE, SITE_AUDIT_LLM_MARGIN_MULTIPLE, VIDEO_ANALYSIS_LLM_MARGIN_MULTIPLE, MEDIA_TRANSCRIBE_MARGIN_MULTIPLE, MEDIA_TRANSCRIBE_HOLD_MC, MC_PER_USD;
+function mediaTranscribeChargeMc(durationSecs) {
+  const mins = Math.max(1, Math.ceil((durationSecs || 60) / 60));
+  return MEDIA_TRANSCRIBE_FLAT_MC + MEDIA_TRANSCRIBE_PER_MIN_MC * mins;
+}
+var MC_COSTS, MC_PER_BROWSER_MS, BROWSER_OPEN_MIN_BALANCE_MC, MC_PER_CREDIT, BILLING_MC_PER_USD_NUMERATOR, BILLING_MC_PER_USD_DENOMINATOR, BILLING_BASIS_POINTS, USD_NANOS, SCHEDULED_RUN_BASE_CREDITS, SCHEDULED_RUN_BASE_MC, SCHEDULED_AGENT_LLM_MARKUP_BPS, SCHEDULED_AGENT_LLM_MARKUP, SCHEDULED_RUN_DEFAULT_MAX_CREDITS, CONNECTED_USAGE_RATE_POLICY_VERSION, CONNECTED_ACTIVE_CONNECTION_MONTHLY_CREDITS, CONNECTED_ACTIVE_CONNECTION_MONTHLY_MC, CONNECTED_FUNCTION_RUN_CREDITS, CONNECTED_PROXY_REQUEST_CREDITS, CONNECTED_COMPUTE_CREDITS_PER_SECOND, CONNECTED_FUNCTION_RUN_MC, CONNECTED_PROXY_REQUEST_MC, CONNECTED_COMPUTE_MC_PER_SECOND, CONNECTED_USAGE_RATE_POLICY, SCHEDULED_ACTIONS_RATE_POLICY, CREDIT_COST_CATALOG, CONCURRENCY_PRICE_ID, SUBSCRIPTION_TIERS, SUBSCRIPTION_TIER_BY_KEY, PRIVATE_OFFERS, CONCURRENCY_SLOT_PRICE_USD, CONCURRENCY_SLOT_PRICE_CURRENCY, CONCURRENCY_SLOT_PRICE_INTERVAL, CONCURRENCY_SLOT_PRICE_LABEL, CONCURRENCY_SLOT_TERMINAL_COMMAND, MEMORY_FREE_QUOTA_BYTES, MEMORY_PLANS, MEMORY_PLAN_QUOTA, MEMORY_MARGIN_MULTIPLE, MEMORY_PRO_COST_BUDGET_USD, LedgerOperation, MEMORY_AI_MARGIN_MULTIPLE, SITE_AUDIT_LLM_MARGIN_MULTIPLE, VIDEO_ANALYSIS_LLM_MARGIN_MULTIPLE, MEDIA_TRANSCRIBE_FLAT_MC, MEDIA_TRANSCRIBE_PER_MIN_MC, MEDIA_TRANSCRIBE_HOLD_MC, MC_PER_USD;
 var init_rates = __esm({
   "src/api/rates.ts"() {
     "use strict";
@@ -9327,9 +9331,9 @@ var init_rates = __esm({
         key: "yt_transcription",
         label: "YouTube transcription",
         aliases: ["youtube_transcribe", "youtube transcript", "transcription", "yt_transcription"],
-        credits: mcToCredits(MC_COSTS.yt_transcription),
-        unit: "per minute",
-        notes: "A 5-minute hold is taken, then reconciled to actual video duration."
+        credits: mcToCredits(2e4),
+        unit: "base + 2/min of video",
+        notes: "Flat 200-credit base plus 2 credits per minute of video, reconciled to the real video length after transcription (e.g. a 10-minute video is about 220 credits)."
       },
       {
         key: "fb_ad",
@@ -9390,17 +9394,17 @@ var init_rates = __esm({
         key: "google_ads_transcribe",
         label: "Google ad video transcription",
         aliases: ["google_ads_transcribe", "google ad transcript"],
-        credits: mcToCredits(MC_COSTS.google_ads_transcribe),
-        unit: "per minute (typical)",
-        notes: "A hold is taken, then reconciled to the real transcription cost for the actual audio duration."
+        credits: mcToCredits(2e4),
+        unit: "base + 2/min of video",
+        notes: "Flat 200-credit base plus 2 credits per minute of video, reconciled to the real video length after transcription (e.g. a 10-minute video is about 220 credits)."
       },
       {
         key: "fb_transcribe",
         label: "Facebook video / ad transcription",
         aliases: ["facebook_transcribe", "facebook_video_transcribe", "fb_transcribe", "fb ad transcript"],
-        credits: mcToCredits(MC_COSTS.fb_transcribe),
-        unit: "per minute (typical)",
-        notes: "A hold is taken, then reconciled to the real transcription cost for the actual audio duration."
+        credits: mcToCredits(2e4),
+        unit: "base + 2/min of video",
+        notes: "Flat 200-credit base plus 2 credits per minute of video, reconciled to the real video length after transcription (e.g. a 10-minute video is about 220 credits)."
       },
       {
         key: "instagram_profile",
@@ -9422,9 +9426,9 @@ var init_rates = __esm({
         key: "instagram_transcribe",
         label: "Instagram media transcription",
         aliases: ["instagram transcript", "instagram reel transcript", "ig transcribe"],
-        credits: mcToCredits(MC_COSTS.instagram_transcribe),
-        unit: "per minute (typical)",
-        notes: "Whisper transcription of selected Instagram audio/video media via fal.ai. A hold is taken, then reconciled to the real transcription cost for the actual audio duration."
+        credits: mcToCredits(2e4),
+        unit: "base + 2/min of video",
+        notes: "Whisper transcription of selected Instagram audio/video media. Flat 200-credit base plus 2 credits per minute of video, reconciled to the real video length after transcription (e.g. a 10-minute video is about 220 credits)."
       },
       {
         key: "browser_minute",
@@ -9603,8 +9607,9 @@ var init_rates = __esm({
     MEMORY_AI_MARGIN_MULTIPLE = 3;
     SITE_AUDIT_LLM_MARGIN_MULTIPLE = 1.5;
     VIDEO_ANALYSIS_LLM_MARGIN_MULTIPLE = 1.5;
-    MEDIA_TRANSCRIBE_MARGIN_MULTIPLE = 3;
-    MEDIA_TRANSCRIBE_HOLD_MC = MC_COSTS.fb_transcribe * 3;
+    MEDIA_TRANSCRIBE_FLAT_MC = 2e4;
+    MEDIA_TRANSCRIBE_PER_MIN_MC = 200;
+    MEDIA_TRANSCRIBE_HOLD_MC = MEDIA_TRANSCRIBE_FLAT_MC + MEDIA_TRANSCRIBE_PER_MIN_MC * 30;
     MC_PER_USD = SUBSCRIPTION_TIERS["price_1TmiHRS8aAcsk3TGwmSNfNIa"].credits_mc / SUBSCRIPTION_TIERS["price_1TmiHRS8aAcsk3TGwmSNfNIa"].monthly_usd;
   }
 });
@@ -16955,8 +16960,7 @@ var init_youtube_routes = __esm({
       const { videoId } = parsed.data;
       const id = videoId.trim();
       const user = c.get("user");
-      const holdMins = 5;
-      const holdMc = MC_COSTS.yt_transcription * holdMins;
+      const holdMc = MEDIA_TRANSCRIBE_HOLD_MC;
       const gate = await acquireConcurrencyGate(user, "youtube_transcribe", {
         reuseLockId: c.req.header("x-mcp-scraper-concurrency-lock"),
         metadata: { videoId: id }
@@ -16972,8 +16976,7 @@ var init_youtube_routes = __esm({
         const chunkSecs = lastChunk && Number.isFinite(lastChunk.timestamp[1]) ? lastChunk.timestamp[1] : 0;
         const durationSecs = Number.isFinite(result.durationMs) ? result.durationMs / 1e3 : 0;
         const videoSecs = Math.max(chunkSecs, durationSecs) || 60;
-        const actualMins = Math.max(1, Math.ceil(videoSecs / 60));
-        const actualMc = MC_COSTS.yt_transcription * actualMins;
+        const actualMc = mediaTranscribeChargeMc(videoSecs);
         const diff = holdMc - actualMc;
         if (diff > 0) await creditMc(user.id, diff, LedgerOperation.TRANSCRIPTION_REFUND, "overestimate refund");
         else if (diff < 0) await debitMc(user.id, -diff, LedgerOperation.TRANSCRIPTION, id);
@@ -18211,7 +18214,6 @@ var init_facebook_ad_routes = __esm({
     import_zod19 = require("zod");
     init_db();
     init_rates();
-    init_cost_rates();
     init_BrowserDriver();
     init_FacebookAdExtractor();
     init_FacebookAdGraphql();
@@ -18369,7 +18371,7 @@ var init_facebook_ad_routes = __esm({
         if (!ok) return c.json(insufficientBalanceResponse(balance_mc, holdMc), 402);
         debited = true;
         const transcript = await transcribeFacebookVideoUrl(videoUrl, "# Facebook Ad Transcript");
-        const settleMc = Math.round(transcript.costUsd * MEDIA_TRANSCRIBE_MARGIN_MULTIPLE * MC_PER_USD);
+        const settleMc = mediaTranscribeChargeMc(transcript.durationMs / 1e3);
         const diff = holdMc - settleMc;
         if (diff > 0) await creditMc(fbUser.id, diff, LedgerOperation.FB_TRANSCRIBE_REFUND, "real-cost reconcile");
         else if (diff < 0) await debitMc(fbUser.id, -diff, LedgerOperation.FB_TRANSCRIBE, videoUrl);
@@ -18420,7 +18422,7 @@ var init_facebook_ad_routes = __esm({
         const video = await extractFacebookOrganicVideoFromPage(page, sourceUrl.href, body.quality);
         const kernelMs = Date.now() - kernelStartMs;
         const transcript = await transcribeFacebookVideoUrl(video.videoUrl, "# Facebook Organic Video Transcript");
-        const settleMc = Math.round((transcript.costUsd + kernelCostUsd(kernelMs, true)) * MEDIA_TRANSCRIBE_MARGIN_MULTIPLE * MC_PER_USD);
+        const settleMc = mediaTranscribeChargeMc(transcript.durationMs / 1e3);
         const diff = holdMc - settleMc;
         if (diff > 0) await creditMc(fbUser.id, diff, LedgerOperation.FB_TRANSCRIBE_REFUND, "real-cost reconcile");
         else if (diff < 0) await debitMc(fbUser.id, -diff, LedgerOperation.FB_TRANSCRIBE, sourceUrl.href);
@@ -18925,7 +18927,7 @@ var init_google_ads_routes = __esm({
         if (!ok) return c.json(insufficientBalanceResponse(balance_mc, holdMc), 402);
         debited = true;
         const transcript = await runWithCostContext({ ...currentCostContext(), op: "google_ads_transcribe", userId: user.id }, () => transcribeMediaUrl(videoUrl, "# Google Ad Transcript"));
-        const settleMc = Math.round(transcript.costUsd * MEDIA_TRANSCRIBE_MARGIN_MULTIPLE * MC_PER_USD);
+        const settleMc = mediaTranscribeChargeMc(transcript.durationMs / 1e3);
         const diff = holdMc - settleMc;
         if (diff > 0) await creditMc(user.id, diff, LedgerOperation.GOOGLE_ADS_TRANSCRIBE_REFUND, "real-cost reconcile");
         else if (diff < 0) await debitMc(user.id, -diff, LedgerOperation.GOOGLE_ADS_TRANSCRIBE, videoUrl);
@@ -19600,7 +19602,7 @@ var init_instagram_routes = __esm({
             if (!tx.ok) return c.json(insufficientBalanceResponse(tx.balance_mc, holdMc), 402);
             transcriptDebited = true;
             transcript = await transcribeMediaUrl(audioUrl, "# Instagram Media Transcript");
-            const settleMc = Math.round(transcript.costUsd * MEDIA_TRANSCRIBE_MARGIN_MULTIPLE * MC_PER_USD);
+            const settleMc = mediaTranscribeChargeMc(transcript.durationMs / 1e3);
             const diff = holdMc - settleMc;
             if (diff > 0) await creditMc(user.id, diff, LedgerOperation.INSTAGRAM_TRANSCRIBE_REFUND, "real-cost reconcile");
             else if (diff < 0) await debitMc(user.id, -diff, LedgerOperation.INSTAGRAM_TRANSCRIBE, sourceUrl.href);
@@ -31109,7 +31111,7 @@ var PACKAGE_VERSION;
 var init_version = __esm({
   "src/version.ts"() {
     "use strict";
-    PACKAGE_VERSION = "0.34.0";
+    PACKAGE_VERSION = "0.34.1";
   }
 });
 
@@ -34557,7 +34559,7 @@ var init_browser_agent_tool_schemas = __esm({
       wait_ms: import_zod37.z.number().int().min(0).max(18e4).optional().describe("How long to wait for the answer stream to finish. Defaults to 90000 when a prompt is sent, 8000 for passive capture."),
       first_party_domain: import_zod37.z.string().optional().describe("The brand/site being researched, e.g. example.com \u2014 sources on this domain are tagged First-party/vendor."),
       reset: import_zod37.z.boolean().default(false).describe("Clear any previously buffered stream for this page before capturing."),
-      export: import_zod37.z.boolean().default(false).describe("Write JSON/CSV/TSV/HTML exports to MCP_SCRAPER_OUTPUT_DIR/fanout, returning relative paths.")
+      export: import_zod37.z.boolean().default(false).describe("When using the installed local MCP server, write JSON/CSV/TSV/HTML exports to MCP_SCRAPER_OUTPUT_DIR/fanout. Hosted clients such as ChatGPT always receive the complete structured result inline and leave exports null.")
     };
     FanoutSourceOutput = import_zod37.z.object({
       url: import_zod37.z.string(),
@@ -34598,6 +34600,7 @@ var init_browser_agent_tool_schemas = __esm({
         byCategory: import_zod37.z.record(import_zod37.z.number().int())
       }).describe("Objective aggregates: top sourced sites by frequency, citation order, and URL-category counts."),
       first_party_domain: import_zod37.z.string().nullable(),
+      export_error: import_zod37.z.string().nullable().optional().describe("Non-fatal local export failure. The inline capture remains complete when this is present."),
       exports: import_zod37.z.object({
         relativeTo: import_zod37.z.string(),
         dir: import_zod37.z.string(),
@@ -35167,6 +35170,7 @@ function registerBrowserAgentMcpTools(server, opts) {
   const baseUrl = opts.baseUrl.replace(/\/$/, "");
   const consoleBase = (opts.consoleBaseUrl ?? opts.baseUrl).replace(/\/$/, "");
   const timeoutMs = opts.timeoutMs ?? 9e4;
+  const savesReportsLocally = opts.savesReportsLocally !== false;
   async function req(method, path6, body, requestTimeoutMs = timeoutMs) {
     try {
       const res = await fetch(`${baseUrl}${path6}`, {
@@ -35818,13 +35822,13 @@ function registerBrowserAgentMcpTools(server, opts) {
     "query_fanout_workflow",
     {
       title: "Capture AI Search Fan-Out",
-      description: "Capture the query fan-out behind a ChatGPT or Claude web-search answer for AEO: sub-queries issued, every researched URL split into cited vs browsed-only, and top sourced sites. Returns raw structured data for you to classify and analyze. Set export=true for JSON/CSV/TSV/HTML artifacts. WRITE NOTE: passing prompt submits a real message in the user's logged-in account \u2014 only send when the user wants that; omit it to capture a prompt the user just ran. The session must already be open on chatgpt.com or claude.ai (see browser_profile_connect) while the prompt streams. NOT for Google AI Overview \u2014 use harvest_paa for that.",
+      description: "Capture the query fan-out behind a ChatGPT or Claude web-search answer for AEO: sub-queries issued, every researched URL split into cited vs browsed-only, and top sourced sites. Complete structured data is always returned inline for analysis. export=true additionally writes JSON/CSV/TSV/HTML only from an installed local MCP server; hosted OAuth/HTTP clients receive exports=null and use the inline data. A local export failure does not discard a successful capture. WRITE NOTE: passing prompt submits a real message in the user's logged-in account \u2014 only send when the user wants that; omit it to capture a prompt the user just ran. The session must already be open on chatgpt.com or claude.ai (see browser_profile_connect) while the prompt streams. NOT for Google AI Overview \u2014 use harvest_paa for that.",
       inputSchema: BrowserCaptureFanoutInputSchema,
       outputSchema: recordOutputSchema("query_fanout_workflow", BrowserCaptureFanoutOutputSchema),
       annotations: annotations("Capture AI Search Fan-Out")
     },
     async (input) => {
-      const emit = (result, exports3) => structuredResult({
+      const emit = (result, exports3, exportError2 = null) => structuredResult({
         ok: true,
         tool: "query_fanout_workflow",
         session_id: input.session_id,
@@ -35841,6 +35845,7 @@ function registerBrowserAgentMcpTools(server, opts) {
         aggregates: result.aggregates,
         first_party_domain: result.firstPartyDomain,
         exports: exports3 ?? null,
+        export_error: exportError2,
         ...result.debug ? { debug: result.debug } : {}
       });
       const res = await req("POST", `/agent/sessions/${input.session_id}/capture-fanout`, {
@@ -35848,26 +35853,20 @@ function registerBrowserAgentMcpTools(server, opts) {
         wait_ms: input.wait_ms,
         first_party_domain: input.first_party_domain,
         reset: input.reset,
-        export: input.export
+        export: false
       }, Math.max(timeoutMs, (input.wait_ms ?? (input.prompt ? 9e4 : 8e3)) + 3e4));
       if (!res.ok) return errorResult("query_fanout_workflow", res.data, input.session_id);
       const hosted = res.data?.result ?? res.data;
-      let exports2 = res.data?.exports ?? null;
-      if (input.export && !exports2 && isEnrichedFanout(hosted)) {
+      let exports2 = null;
+      let exportError = null;
+      if (input.export && savesReportsLocally && isEnrichedFanout(hosted)) {
         try {
           exports2 = exportFanout(hosted);
         } catch (err) {
-          return errorResult(
-            "query_fanout_workflow",
-            {
-              error: `Fan-out captured but local export failed: ${err instanceof Error ? err.message : String(err)}`,
-              result: hosted
-            },
-            input.session_id
-          );
+          exportError = `Fan-out captured successfully, but local export failed: ${err instanceof Error ? err.message : String(err)}`;
         }
       }
-      return emit(hosted, exports2);
+      return emit(hosted, exports2, exportError);
     }
   );
 }
@@ -41571,6 +41570,19 @@ var init_browser_agent_console = __esm({
 });
 
 // src/api/stripe-routes.ts
+async function cancelStandaloneMemorySub(memSubId, baseSubId) {
+  if (!memSubId || memSubId === baseSubId) return;
+  try {
+    const memSub = await stripe.subscriptions.retrieve(memSubId);
+    const priceId = memSub.items.data[0]?.price?.id;
+    const stillOpen = memSub.status !== "canceled" && !memSub.cancel_at_period_end;
+    if (priceId && priceId in MEMORY_PLANS && stillOpen) {
+      await stripe.subscriptions.update(memSubId, { cancel_at_period_end: true });
+    }
+  } catch (err) {
+    console.error("[stripe] could not wind down standalone memory subscription", memSubId, err instanceof Error ? err.message : err);
+  }
+}
 function linePriceId(line) {
   const l = line;
   return l?.price?.id ?? l?.pricing?.price_details?.price ?? l?.plan?.id;
@@ -41625,14 +41637,7 @@ var init_stripe_routes = __esm({
               await creditMc(user.id, tier.credits_mc, LedgerOperation.SUBSCRIPTION, `${tier.label} subscription credits`, invoice.id, { neverExpires: tier.credits_never_expire });
               await setSubscriptionTier(user.id, tier.tier, tier.concurrency, subId ?? user.subscription_id);
               if (tier.includes_memory) {
-                const standaloneMemorySub = user.memory_subscription_id;
-                if (standaloneMemorySub) {
-                  try {
-                    await stripe.subscriptions.update(standaloneMemorySub, { cancel_at_period_end: true });
-                  } catch (err) {
-                    console.error("[stripe] could not wind down standalone memory subscription", standaloneMemorySub, err instanceof Error ? err.message : err);
-                  }
-                }
+                await cancelStandaloneMemorySub(user.memory_subscription_id, subId ?? user.subscription_id);
                 await setMemoryPlan(user.id, "pro", null);
                 await syncMemoryKeyPlan(user, resolveEffectiveMemoryPlan({ ...user, memory_plan: "pro", subscription_tier: tier.tier }));
               }
@@ -41684,6 +41689,7 @@ var init_stripe_routes = __esm({
           await setSubscriptionTier(user.id, newTier, live ? tier.concurrency : 0, live ? sub.id : null);
           if (tier.includes_memory) {
             const newMemoryPlan = live ? "pro" : "free";
+            if (live) await cancelStandaloneMemorySub(user.memory_subscription_id, sub.id);
             await setMemoryPlan(user.id, newMemoryPlan, live ? sub.id : null);
             await syncMemoryKeyPlan(user, resolveEffectiveMemoryPlan({ ...user, memory_plan: newMemoryPlan }));
           } else if (user.memory_subscription_id === sub.id) {
@@ -50129,15 +50135,25 @@ var init_server = __esm({
           customerId = customer.id;
           await setStripeCustomerId(user.id, customerId);
         }
-        if (user.subscription_id) {
-          const sub = await stripeClient.subscriptions.retrieve(user.subscription_id);
+        let existingSubId = user.subscription_id;
+        if (!existingSubId) {
+          try {
+            const active = await stripeClient.subscriptions.list({ customer: customerId, status: "active", limit: 20 });
+            const baseSub = active.data.find((s) => findBasePlanItem(s));
+            if (baseSub) existingSubId = baseSub.id;
+          } catch (err) {
+            console.warn("[billing/subscribe] could not check for existing subscription", err instanceof Error ? err.message : err);
+          }
+        }
+        if (existingSubId) {
+          const sub = await stripeClient.subscriptions.retrieve(existingSubId);
           const itemId = findBasePlanItem(sub)?.id;
           if (itemId) {
-            await stripeClient.subscriptions.update(user.subscription_id, {
+            await stripeClient.subscriptions.update(existingSubId, {
               items: [{ id: itemId, price: tier.price_id }],
               proration_behavior: "create_prorations"
             });
-            const billed = await chargeTierChangeNow(stripeClient, user.subscription_id, customerId);
+            const billed = await chargeTierChangeNow(stripeClient, existingSubId, customerId);
             if (!billed.ok) {
               return c.json({
                 updated: true,
@@ -50255,15 +50271,25 @@ var init_server = __esm({
           customerId = customer.id;
           await setStripeCustomerId(user.id, customerId);
         }
-        if (user.subscription_id) {
-          const sub = await stripeClient.subscriptions.retrieve(user.subscription_id);
+        let existingSubId = user.subscription_id;
+        if (!existingSubId) {
+          try {
+            const active = await stripeClient.subscriptions.list({ customer: customerId, status: "active", limit: 20 });
+            const baseSub = active.data.find((s) => findBasePlanItem(s));
+            if (baseSub) existingSubId = baseSub.id;
+          } catch (err) {
+            console.warn("[billing/subscribe] could not check for existing subscription", err instanceof Error ? err.message : err);
+          }
+        }
+        if (existingSubId) {
+          const sub = await stripeClient.subscriptions.retrieve(existingSubId);
           const itemId = findBasePlanItem(sub)?.id;
           if (itemId) {
-            await stripeClient.subscriptions.update(user.subscription_id, {
+            await stripeClient.subscriptions.update(existingSubId, {
               items: [{ id: itemId, price: tier.price_id }],
               proration_behavior: "create_prorations"
             });
-            const billed = await chargeTierChangeNow(stripeClient, user.subscription_id, customerId);
+            const billed = await chargeTierChangeNow(stripeClient, existingSubId, customerId);
             if (!billed.ok) {
               return c.json({
                 updated: true,