@kolbo/mcp 1.113.0 → 1.113.2

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@kolbo/mcp",
-  "version": "1.113.0",
+  "version": "1.113.2",
   "description": "Kolbo AI MCP Server - Generate images, videos, music, speech, and sound effects from Claude Code",
   "main": "src/index.js",
   "bin": {
@@ -11,7 +11,7 @@
     "smoke": "node scripts/smoke.js",
     "check-skill-bundle": "node scripts/check-skill-bundle.js",
     "check-parity": "node scripts/check-parity.js",
-    "prepublishOnly": "node --test test/flow-tools.test.js && node --test test/widget-resource-versions.test.js && node scripts/smoke.js && node scripts/check-tool-contract.js && node --test test/blender-tools.test.js && node scripts/check-skill-bundle.js && node scripts/check-parity.js && node scripts/check-widget-fields.js && node scripts/check-widget-render.js && node scripts/check-model-catalog.js && node scripts/check-skill-tools.js && node scripts/check-doctrine-parity.js && node scripts/check-submission-contract.js && node scripts/check-install.js",
+    "prepublishOnly": "node --test test/flow-tools.test.js && node --test test/widget-resource-versions.test.js test/reference-preview-inline.test.js && node scripts/smoke.js && node scripts/check-tool-contract.js && node --test test/blender-tools.test.js && node scripts/check-skill-bundle.js && node scripts/check-parity.js && node scripts/check-widget-fields.js && node scripts/check-widget-render.js && node scripts/check-model-catalog.js && node scripts/check-skill-tools.js && node scripts/check-doctrine-parity.js && node scripts/check-submission-contract.js && node scripts/check-install.js",
     "check-model-catalog": "node scripts/check-model-catalog.js",
     "check-widget-fields": "node scripts/check-widget-fields.js",
     "check-widget-render": "node scripts/check-widget-render.js",

package/src/apps/html.js

@@ -225,7 +225,11 @@ document.addEventListener('click', function (e) {
     list.push(it);
   });
   var at = peekItem(node);
-  openPeek(at.url, at.kind, at.cap, { list: list, time: +node.getAttribute('data-peek-time') || 0 });
+  openPeek(at.url, at.kind, at.cap, {
+    list: list,
+    time: +node.getAttribute('data-peek-time') || 0,
+    inline: node.getAttribute('data-peek-inline') === 'true'
+  });
 }, true);
 var peekList = [];
 var peekIndex = 0;
@@ -298,7 +302,9 @@ function openPeek(url, kind, cap, opts) {
   box.hidden = false;
   document.documentElement.classList.add('k-peek-open');
   if (window.kolbo && window.kolbo.notifySize) window.kolbo.notifySize();
-  if (peekWentFullscreen || !window.kolbo || !window.kolbo.requestDisplayMode) return;
+  // Reference previews stay in the chat card instead of handing placement to
+  // the host's fullscreen surface. Result previews keep their existing expand.
+  if (opts.inline || peekWentFullscreen || !window.kolbo || !window.kolbo.requestDisplayMode) return;
   try {
     window.kolbo.requestDisplayMode('fullscreen').then(function (res) {
       if (!(res && res.mode === 'fullscreen')) return;

package/src/apps/index.js

@@ -366,6 +366,9 @@ async function modelInfoMap(client) {
 const DISPLAY_ALIASES = {
   'flux-3-draft-enhance': { base: 'seedance-2-5', name: 'Seedance 2.5 Draft Enhance' },
   'video-draft-enhance': { base: 'seedance-2-5', name: 'Seedance 2.5 Draft Enhance' },
+  // Billing SKU every ElevenLabs CLONED voice reports; its version shows in the
+  // separate "Model V4/V3/V2" chip, so the model chip is just the provider.
+  'fal_elevenlabs_tts': { base: 'eleven_v4', name: 'ElevenLabs' },
 };
 // Operations whose engine is picked server-side when no model is passed.
 const OPERATION_ENGINES = { draft_enhance: 'flux-3-draft-enhance' };
@@ -530,10 +533,16 @@ async function resolveCatalogAspectRatio(client, modelId, requested, type) {
 
 // Same shape and TTL as the model catalog above: the voice catalog is stable,
 // and a per-generation /v1/voices round trip would be paid on every speech card.
-const voiceCache = new Map(); // apiBase → { at, byKey }
+// Keyed per caller, not per server: /v1/voices includes the caller's own cloned
+// voices, so one shared entry (the remote connector serves every user from one
+// process) left everyone else's clones unresolved for the whole TTL.
+const voiceCache = new Map(); // apiBase + key hash → { at, byKey }
 
 async function voiceInfoMap(client) {
-  const cacheKey = client.apiBase || 'default';
+  const keyHash = client.apiKey
+    ? crypto.createHash('sha256').update(String(client.apiKey)).digest('hex').slice(0, 16)
+    : 'anon';
+  const cacheKey = `${client.apiBase || 'default'}|${keyHash}`;
   const hit = voiceCache.get(cacheKey);
   if (hit && Date.now() - hit.at < ICON_TTL_MS) return hit.byKey;
   const byKey = new Map();

package/src/apps/widgets/generation.js

@@ -423,7 +423,7 @@ function referenceHTML(sc) {
   for (var i = 0; i < refs.length; i++) {
     var url = esc(refs[i].url);
     var title = 'Reference ' + refs[i].kind + ' ' + (i + 1) + ' of ' + refs.length;
-    var peek = ' data-peek="' + url + '" data-peek-kind="' + refs[i].kind + '" data-peek-cap="' + title + '"';
+    var peek = ' data-peek="' + url + '" data-peek-kind="' + refs[i].kind + '" data-peek-cap="' + title + '" data-peek-inline="true"';
     if (refs[i].kind === 'video') {
       // #t=0.1 so the poster is a real frame, not a black canvas.
       h += '<video class="k-ref-thumb k-peek-hit" src="' + url + '#t=0.1" muted playsinline preload="metadata" title="'

package/src/tools/generate.js

@@ -1201,7 +1201,8 @@ function registerGenerateTools(server, client, options = {}) {
       // the card was labelling those "Smart Select", which is not even a text-to-
       // speech option, and naming the voice the caller typed rather than the one
       // that spoke. Same resolution addDisplayNames does for the polling path.
-      const ranVoice = (await voiceInfo(client, result.result?.voice).catch(() => null)) || voiceRecord;
+      const ranVoice = statusVoice(result.result)
+        || (await voiceInfo(client, result.result?.voice).catch(() => null)) || voiceRecord;
       return uiCompleted({
         tool: 'generate_speech', kind: 'audio', gen, client, prompt: text,
         model: result.result?.model || model,
@@ -1314,13 +1315,21 @@ function registerGenerateTools(server, client, options = {}) {
       r.model_name = (info && info.name) || r.model;
       r.model_icon = (info && info.icon) || null;
     }
-    if (r.voice) {
+    // The API names a cloned voice itself (voice_name/voice_thumbnail): clones
+    // are not reliably in the cached catalog, so keep the API's answer.
+    if (r.voice && !statusVoice(r)) {
       const v = await voiceInfo(client, r.voice).catch(() => null);
       r.voice_name = (v && v.name) || r.voice;
       r.voice_thumbnail = (v && v.thumbnail) || null;
     }
   }
 
+  function statusVoice(r) {
+    if (!r || !r.voice_name) return null;
+    const id = /^custom_/i.test(String(r.requested_voice_id || '')) ? r.requested_voice_id : r.voice;
+    return { id, name: r.voice_name, thumbnail: r.voice_thumbnail || null };
+  }
+
   // ─── get_generation_status ─────────────────────────────────
   server.tool(
     'get_generation_status',