ayphic-mcp-server 1.2.0 → 1.2.1

package/index.js

@@ -375,13 +375,13 @@ const MCP_TOOLS = [
   },
   {
     name: 'get_transcript',
-    description: 'REQUIRED FIRST STEP FOR VIDEO CAPTIONS: Extract the authentic speech transcript with verbatim word-level timestamps (seconds) from the video/audio clip. When the user asks to add captions/subtitles to a video, you MUST call this tool first to obtain the authentic spoken words and timings from the video before calling captions.',
+    description: 'Inspect speech text and retrieve authentic word-level timestamps (seconds) from video/audio. Use this tool to read what was spoken, inspect timestamps, find filler words (ums, ahs) to cut with split_item, or detect pauses for silence removal. (NOTE: For adding captions directly to the timeline, call `captions` directly).',
     inputSchema: {
       type: 'object',
       properties: {
         itemId: {
           type: 'string',
-          description: 'Optional ID of the clip to transcribe (e.g. "clip2"). If omitted, automatically targets the main video.'
+          description: 'Optional ID of the clip to inspect (e.g. "clip2"). If omitted, automatically targets the main video on track V1.'
         },
         mediaUrl: {
           type: 'string',
@@ -392,23 +392,35 @@ const MCP_TOOLS = [
   },
   {
     name: 'captions',
-    description: 'Add, style, or remove animated captions/subtitles across the video timeline. DIRECT DATA TOOL: When adding captions for a video clip, you MUST first call `get_transcript` to get the real speech timestamps, then pass the `transcription` object directly here. You can also customize `template` ("beast", "hustle", "pop", "soft-ai", "karaoke"), or provide custom Remotion JSX `code`, or use `action: "delete"` to remove captions.',
+    description: 'DIRECT CAPTIONS TOOL: Add, style, generate, or remove animated captions/subtitles across the video timeline in one single step. Call this tool directly (e.g. `template: "beast"` or `template: "hustle"`). It automatically extracts audio from the timeline/clip, transcribes speech with word-level timestamps, and places stylized captions on track C1. Presets: "beast", "hustle", "viral-stack", "grape", "poppin", "soft-ai", "gaming-stream", "simple-one-word", "pop", "karaoke", "podcast", "custom". Use `action: "delete"` to remove all captions.',
     inputSchema: {
       type: 'object',
       properties: {
-        transcription: {
-          type: 'object',
-          description: 'The authentic transcription object returned from get_transcript containing { text, words, segments, lines, captions }. Pass this directly from get_transcript.'
-        },
         template: {
           type: 'string',
-          enum: ['viral-stack', 'viral-stacked', 'stacked', 'beast', 'hustle', 'pop', 'soft-ai', 'gaming-stream', 'karaoke', 'grape', 'poppin', 'kinetic-01', 'kinetic-02', 'aarit', 'podcast'],
-          description: 'Visual subtitle preset theme. Recommended: "viral-stack" (stacked kinetic typography with dynamic sizing and slide-up), "beast" (MrBeast style bold uppercase highlight), "hustle" (green glow), "pop" (spring bounce).'
+          enum: ['viral-stack', 'viral-stacked', 'stacked', 'beast', 'hustle', 'pop', 'soft-ai', 'gaming-stream', 'karaoke', 'grape', 'poppin', 'kinetic-01', 'kinetic-02', 'aarit', 'podcast', 'custom'],
+          description: 'Visual subtitle preset theme. Recommended: "beast" (MrBeast bold uppercase highlight), "hustle" (green glow), "viral-stack" (stacked kinetic typography), "pop" (spring bounce).'
+        },
+        itemId: {
+          type: 'string',
+          description: 'Optional ID of specific clip to transcribe. If omitted, transcribes the main video on track V1.'
+        },
+        trackId: {
+          type: 'string',
+          description: 'Optional track ID (e.g. "V1"). Default is "V1".'
+        },
+        text: {
+          type: 'string',
+          description: 'Optional explicit custom quote text if provided by user in prompt. If transcribing spoken video, leave blank so audio is automatically transcribed.'
+        },
+        transcription: {
+          type: 'object',
+          description: 'Optional direct transcription object if already available containing { text, words, segments }.'
         },
         action: {
           type: 'string',
           enum: ['generate', 'delete', 'remove', 'clear'],
-          description: 'Action to perform. Use "generate" to add/update captions, or "delete"/"remove"/"clear" to remove all captions from the timeline.'
+          description: 'Action to perform. Default is "generate". Use "delete"/"remove"/"clear" to remove all captions from the timeline.'
         },
         code: {
           type: 'string',
@@ -417,12 +429,12 @@ const MCP_TOOLS = [
         words: {
           type: 'array',
           items: { type: 'object' },
-          description: 'Array of word-level timing objects [{ text: string, start: number, end: number }] from get_transcript.'
+          description: 'Optional array of word-level timing objects [{ text: string, start: number, end: number }].'
         },
         captions: {
           type: 'array',
           items: { type: 'object' },
-          description: 'Array of timed caption segments [{ text: string, start: number, end: number, words: array }] from get_transcript.'
+          description: 'Optional array of timed caption segments [{ text: string, start: number, end: number, words: array }].'
         },
         fontSize: {
           type: ['number', 'string'],
@@ -622,33 +634,13 @@ async function executeAyphicTool(toolName, params) {
       };
     }
 
-    // For get_timeline, return the structured timeline data
-    if (resolvedToolName === 'get_timeline' || resolvedToolName === 'getTimeline') {
-      return {
-        isError: false,
-        content: [
-          {
-            type: 'text',
-            text: JSON.stringify(data, null, 2)
-          }
-        ]
-      };
-    }
-
-    // For all action/editing tools, return concise summary
-    const summaryMsg = data.message || `Successfully executed ${cleanName}`;
-    const resultObj = {
-      success: true,
-      message: summaryMsg,
-      ...(data.element ? { element: data.element } : {})
-    };
-
+    // Return the full structured result from data (including transcripts, bounding boxes, overlaps, elements, etc.)
     return {
       isError: false,
       content: [
         {
           type: 'text',
-          text: JSON.stringify(resultObj, null, 2)
+          text: JSON.stringify(data, null, 2)
         }
       ]
     };

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "ayphic-mcp-server",
-  "version": "1.2.0",
+  "version": "1.2.1",
   "description": "Model Context Protocol (MCP) Server for Ayphic AI Video Editor",
   "main": "index.js",
   "type": "module",