get_meeting_transcript

Instructions

Implementation Reference

• granola_mcp_server/server.py:549-574 (handler)

  The _get_meeting_transcript method that executes the tool logic for retrieving a meeting transcript, formats it with speaker info, language, confidence, and content.

  async def _get_meeting_transcript(self, meeting_id: str) -> List[TextContent]:
      """Get meeting transcript."""
      if not self.cache_data:
          return [TextContent(type="text", text="No meeting data available")]
      
      if meeting_id not in self.cache_data.transcripts:
          return [TextContent(type="text", text=f"No transcript available for meeting '{meeting_id}'")]
      
      transcript = self.cache_data.transcripts[meeting_id]
      meeting = self.cache_data.meetings.get(meeting_id)
      
      output = [f"# Transcript: {meeting.title if meeting else meeting_id}\n"]
      
      if transcript.speakers:
          output.append(f"**Speakers:** {', '.join(transcript.speakers)}")
      
      if transcript.language:
          output.append(f"**Language:** {transcript.language}")
      
      if transcript.confidence:
          output.append(f"**Confidence:** {transcript.confidence:.2%}")
      
      output.append("\n## Transcript Content\n")
      output.append(transcript.content)
      
      return [TextContent(type="text", text="\n".join(output))]

• granola_mcp_server/server.py:141-154 (registration)

  Registration of the get_meeting_transcript tool in the list_tools() handler with input schema requiring meeting_id.

  Tool(
      name="get_meeting_transcript",
      description="Get transcript for a specific meeting",
      inputSchema={
          "type": "object",
          "properties": {
              "meeting_id": {
                  "type": "string", 
                  "description": "Meeting ID to get transcript for"
              }
          },
          "required": ["meeting_id"]
      }
  ),

• granola_mcp_server/server.py:206-207 (handler)

  The call_tool dispatcher that routes the 'get_meeting_transcript' tool name to the _get_meeting_transcript handler method.

  elif name == "get_meeting_transcript":
      return await self._get_meeting_transcript(arguments["meeting_id"])

• granola_mcp_server/models.py:30-36 (schema)

  The MeetingTranscript Pydantic model used as the schema for transcript data, containing meeting_id, content, speakers, language, and confidence fields.

  class MeetingTranscript(BaseModel):
      """Meeting transcript information."""
      meeting_id: str
      content: str
      speakers: List[str] = []
      language: Optional[str] = None
      confidence: Optional[float] = None

• granola_mcp_server/models.py:39-43 (helper)

  The CacheData model that stores transcripts in a dictionary keyed by meeting_id, used by the handler to look up transcripts.

  class CacheData(BaseModel):
      """Complete cache data structure."""
      meetings: Dict[str, MeetingMetadata] = {}
      documents: Dict[str, MeetingDocument] = {}
      transcripts: Dict[str, MeetingTranscript] = {}