get_meeting_transcript

Retrieve meeting transcripts from video calls using bot IDs to access recorded conversations and discussions.

Instructions

Get the transcript from a meeting bot

Input Schema

TableJSON Schema

Name	Required	Description	Default
`bot_id`	Yes	ID of the bot whose transcript to retrieve

Implementation Reference

src/index.ts:489-506 (handler)

The primary handler function that executes the tool logic: validates bot_id input, fetches transcript data from the API endpoint `/api/v1/bots/{bot_id}/transcript`, formats it, and returns as MCP text content.

private async getMeetingTranscript(args: Record<string, unknown>) {
  const bot_id = args.bot_id as string;
  
  if (!bot_id || typeof bot_id !== 'string') {
    throw new Error("Missing or invalid required parameter: bot_id");
  }
  
  const data = await this.makeApiRequest(`/api/v1/bots/${bot_id}/transcript`);

  return {
    content: [
      {
        type: "text",
        text: this.formatTranscriptResponse(data, bot_id),
      },
    ],
  };
}

src/index.ts:238-251 (registration)

Registers the 'get_meeting_transcript' tool in the ListTools response, including its name, description, and input schema definition.

{
  name: "get_meeting_transcript",
  description: "Get the transcript from a meeting bot",
  inputSchema: {
    type: "object",
    properties: {
      bot_id: {
        type: "string",
        description: "ID of the bot whose transcript to retrieve",
      },
    },
    required: ["bot_id"],
  },
},

src/index.ts:241-250 (schema)

Defines the input schema for the tool, specifying the required 'bot_id' string parameter.

inputSchema: {
  type: "object",
  properties: {
    bot_id: {
      type: "string",
      description: "ID of the bot whose transcript to retrieve",
    },
  },
  required: ["bot_id"],
},

src/index.ts:142-183 (helper)

Helper function that formats the raw transcript data from the API into a user-friendly string output, handling both array and object formats, with timestamps and speaker names.

private formatTranscriptResponse(data: any, botId: string): string {
  // If data is an array, it means we got the transcript entries directly
  if (Array.isArray(data)) {
    if (data.length === 0) {
      return `❌ No transcript available for bot ${botId}`;
    }
    
    let transcript = `📝 Meeting Transcript for bot ${botId}:\n\n`;
    transcript += "─".repeat(50) + "\n";
    
    data.forEach((entry: any) => {
      const timestamp = entry.timestamp_ms / 1000;
      const minutes = Math.floor(timestamp / 60);
      const seconds = Math.floor(timestamp % 60);
      const timeStr = `${minutes.toString().padStart(2, '0')}:${seconds.toString().padStart(2, '0')}`;
      
      transcript += `[${timeStr}] ${entry.speaker_name}:\n${entry.transcription}\n\n`;
    });
    
    transcript += "─".repeat(50) + `\n📊 Total entries: ${data.length}`;
    return transcript;
  }
  
  // Handle object response (legacy format)
  if (data.ready && data.transcript) {
    return [
      `📝 Meeting Transcript for bot ${botId}:`,
      "",
      "─".repeat(50),
      data.transcript,
      "─".repeat(50),
    ].join("\n");
  } else {
    const stateIcon = data.transcription_state === "in_progress" ? "🔄" : "⏳";
    return [
      `${stateIcon} Transcript not ready for bot ${botId}`,
      `Current transcription state: ${data.transcription_state}`,
      "",
      "💡 The transcript will be available after the meeting ends and processing completes.",
    ].join("\n");
  }
}

Attendee MCP Server