MCP Tools Reference

The MCP Server exposes 10 high-level tools for Claude Desktop and other MCP clients, which internally use 28 primitive functions for data access and processing.

Table of Contents

• MCP Tools (10) - Tools exposed to MCP clients
• Primitive Functions (28) - Internal implementation details

MCP Tools

These are the tools available when connecting via MCP (Claude Desktop, HTTP/SSE transport).

natural_query (Recommended)

The recommended entry point. Routes your question to the optimal handler automatically.

"Show my rooms"
"Summarize my last meeting"
"Create a meeting for tomorrow at 2pm"
"Who do I collaborate with most?"

Parameters:

query_data

Query meeting data using natural language. Translates your question into database queries.

Parameters:

Examples:

• "Show my meetings this week"
• "How many rooms have I used?"
• "List recordings from January"

agent_query

AI agent for complex queries requiring multi-step reasoning.

Parameters:

Examples:

• "Find meetings with John and summarize what we discussed"
• "Schedule a follow-up with everyone from yesterday's standup"

summarize_meeting

Get an AI-powered summary of a meeting.

Parameters:

query_meeting_context

Analyze meeting mood, sentiment, and engagement.

Parameters:

search_content

Search through transcriptions, chat messages, and documents using semantic similarity.

Parameters:

start_meeting

Start an instant meeting or create a persistent room.

Parameters:

create_room

Create a new persistent room.

Parameters:

create_schedule

Create a new scheduled meeting.

Parameters:

cancel_schedule

Cancel a scheduled meeting.

Parameters:

Primitive Functions

The 28 primitive functions are internal building blocks used by the MCP tools. They are organized into five categories.

Common Parameters

All primitives accept these parameters internally:

Data Query Tools

query_rooms

Query meeting room sessions with filtering and pagination.

Parameters:

Response:

{
  success: true,
  rooms: [
    {
      _id: 'room_id',
      name: 'Weekly Standup',
      owner: 'user@example.com',
      participants: ['user1@example.com', 'user2@example.com'],
      startTime: '2026-01-10T10:00:00Z',
      endTime: '2026-01-10T10:30:00Z',
      hasRecording: true,
      hasTranscription: true
    }
  ],
  total: 42,
  page: 1,
  limit: 20
}

Example:

const result = await toolHandlers.query_rooms({
  limit: 10,
  startDate: '2026-01-01',
  endDate: '2026-01-11',
  apiKey: userContext.apiKey,
  userId: userContext.userId,
  _userContext: securityContext
});

query_schedules

Query scheduled meetings.

Parameters:

Response:

{
  success: true,
  schedules: [
    {
      _id: 'schedule_id',
      title: 'Project Review',
      description: 'Quarterly review meeting',
      scheduledStart: '2026-01-15T14:00:00Z',
      scheduledEnd: '2026-01-15T15:00:00Z',
      organizer: 'user@example.com',
      attendees: ['attendee1@example.com', 'attendee2@example.com'],
      roomName: 'project-review-q1',
      status: 'pending',
      recurring: false
    }
  ],
  total: 15
}

query_stored_rooms

Query saved/persistent rooms.

Parameters:

Response:

{
  success: true,
  storedRooms: [
    {
      _id: 'stored_room_id',
      name: 'team-standup',
      displayName: 'Team Standup',
      owner: 'user@example.com',
      createdAt: '2025-06-01T00:00:00Z',
      lastUsed: '2026-01-10T10:00:00Z',
      settings: {
        autoRecord: true,
        autoTranscribe: true,
        maxParticipants: 20
      }
    }
  ],
  total: 5
}

query_files

Query uploaded files and documents.

Parameters:

Response:

{
  success: true,
  files: [
    {
      _id: 'file_id',
      filename: 'presentation.pdf',
      mimeType: 'application/pdf',
      size: 2048576,
      uploadedBy: 'user@example.com',
      uploadedAt: '2026-01-10T09:30:00Z',
      roomSession: 'room_123',
      url: 'https://s3.../file.pdf'
    }
  ],
  total: 25
}

query_recordings

Query meeting recordings.

Parameters:

Response:

{
  success: true,
  recordings: [
    {
      _id: 'recording_id',
      roomSession: 'room_123',
      roomName: 'Weekly Standup',
      duration: 1800,
      fileSize: 104857600,
      hasTranscription: true,
      transcriptionId: 'trans_456',
      createdAt: '2026-01-10T10:30:00Z',
      url: 'https://s3.../recording.mp4'
    }
  ],
  total: 100
}

get_transcript

Get transcript metadata and content for a recording.

Parameters:

*Either recordingId or transcriptionId is required.

Response:

{
  success: true,
  transcript: {
    _id: 'trans_456',
    recordingId: 'recording_id',
    roomName: 'Weekly Standup',
    duration: 1800,
    speakers: ['user1@example.com', 'user2@example.com'],
    wordCount: 5432,
    createdAt: '2026-01-10T10:30:00Z',
    content: [  // if includeContent: true
      {
        speaker: 'user1@example.com',
        text: 'Good morning everyone...',
        timestamp: 0
      }
    ]
  },
  actionType: 'VIEW_TRANSCRIPT'
}

open_file

Open a file for viewing.

Parameters:

Response:

{
  success: true,
  file: {
    _id: 'file_id',
    filename: 'presentation.pdf',
    mimeType: 'application/pdf',
    url: 'https://s3.../signed-url',
    expiresAt: '2026-01-11T10:00:00Z'
  },
  actionType: 'VIEW_FILE'
}

summarize_transcript

Generate a summary of a meeting transcript.

Parameters:

*Either recordingId or transcriptionId is required.

Summary Types:

• comprehensive - Full meeting summary with key points
• brief - Short executive summary
• action_items - Focus on action items and decisions
• topics - Topic-based summary

Response:

{
  success: true,
  summary: {
    type: 'comprehensive',
    content: '## Meeting Summary\n\nThe team discussed...',
    keyPoints: [
      'Agreed on Q1 priorities',
      'Budget approved for new hire'
    ],
    actionItems: [
      { task: 'Send proposal', assignee: 'John', deadline: '2026-01-15' }
    ],
    participants: ['user1@example.com', 'user2@example.com'],
    duration: 1800
  }
}

query_meeting_context

Query meeting sentiment, mood, and engagement data.

Parameters:

Response:

{
  success: true,
  context: {
    roomSession: 'room_123',
    overallSentiment: 'positive',
    sentimentScore: 0.72,
    engagementLevel: 'high',
    participationBalance: 0.85,
    keyMoments: [
      { timestamp: 300, type: 'decision', description: 'Budget approved' },
      { timestamp: 900, type: 'action_item', description: 'Follow-up scheduled' }
    ],
    timeline: [  // if includeTimeline: true
      { timestamp: 0, sentiment: 0.65 },
      { timestamp: 60, sentiment: 0.70 }
    ]
  }
}

Search Tools

search_content

Semantic search across all content (transcripts, files, messages).

Parameters:

Content Types:

• transcript - Meeting transcripts
• file - Uploaded files
• message - Chat messages

Response:

{
  success: true,
  results: [
    {
      type: 'transcript',
      id: 'trans_456',
      roomName: 'Project Review',
      snippet: '...discussed the budget allocation for Q1...',
      relevance: 0.92,
      timestamp: '2026-01-10T10:15:00Z',
      speaker: 'user@example.com'
    },
    {
      type: 'file',
      id: 'file_789',
      filename: 'Q1-Budget.xlsx',
      snippet: '...quarterly budget projections...',
      relevance: 0.85
    }
  ],
  total: 15,
  query: 'budget allocation'
}

Example:

const result = await toolHandlers.search_content({
  query: 'project timeline discussion',
  limit: 5,
  contentTypes: ['transcript'],
  startDate: '2026-01-01',
  apiKey: userContext.apiKey,
  userId: userContext.userId,
  _userContext: securityContext
});

Graph Analysis Tools

find_collaborators

Find people the user collaborates with most.

Parameters:

Response:

{
  success: true,
  collaborators: [
    {
      email: 'colleague@example.com',
      name: 'John Smith',
      meetingCount: 45,
      totalDuration: 81000,
      lastMeeting: '2026-01-10T10:00:00Z',
      commonTopics: ['project-alpha', 'budget', 'planning']
    }
  ],
  total: 32
}

get_person_meetings

Get all meetings involving a specific person.

Parameters:

Response:

{
  success: true,
  person: {
    email: 'colleague@example.com',
    name: 'John Smith'
  },
  meetings: [
    {
      roomName: 'Weekly Standup',
      date: '2026-01-10T10:00:00Z',
      duration: 1800,
      participants: ['user@example.com', 'colleague@example.com'],
      hasRecording: true
    }
  ],
  total: 45
}

get_graph_stats

Get network statistics for the user.

Parameters:

Response:

{
  success: true,
  stats: {
    totalCollaborators: 32,
    totalMeetings: 271,
    totalDuration: 489600,
    avgMeetingDuration: 1806,
    topCollaborators: [
      { email: 'john@example.com', meetings: 45 }
    ],
    topTopics: [
      { topic: 'project-alpha', mentions: 156 }
    ],
    meetingsByMonth: [
      { month: '2026-01', count: 42 }
    ]
  }
}

find_related_topics

Find topics related to a given topic.

Parameters:

Response:

{
  success: true,
  topic: 'project-alpha',
  relatedTopics: [
    {
      topic: 'budget-planning',
      relevance: 0.85,
      coOccurrences: 23,
      lastMentioned: '2026-01-10T10:00:00Z'
    }
  ]
}

get_meeting_topics

Get topics discussed in a meeting.

Parameters:

Response:

{
  success: true,
  roomSession: 'room_123',
  topics: [
    {
      topic: 'quarterly-review',
      confidence: 0.95,
      mentions: 12,
      speakers: ['user1@example.com', 'user2@example.com']
    }
  ]
}

find_similar_topics

Find semantically similar topics.

Parameters:

Response:

{
  success: true,
  topic: 'budget planning',
  similarTopics: [
    {
      topic: 'financial forecasting',
      similarity: 0.89
    },
    {
      topic: 'resource allocation',
      similarity: 0.76
    }
  ]
}

get_network_stats_graph

Get network statistics with graph visualization data.

Parameters:

Response:

{
  success: true,
  stats: { /* same as get_graph_stats */ },
  graph: {
    nodes: [
      { id: 'user@example.com', label: 'You', type: 'user' },
      { id: 'john@example.com', label: 'John', type: 'collaborator' }
    ],
    edges: [
      { source: 'user@example.com', target: 'john@example.com', weight: 45 }
    ]
  },
  actionType: 'VIEW_GRAPH'
}

get_collaborator_graph

Get graph data focused on collaborator relationships.

Parameters:

Response:

{
  success: true,
  graph: {
    center: 'user@example.com',
    nodes: [
      { id: 'user@example.com', type: 'center', meetings: 271 },
      { id: 'john@example.com', type: 'direct', meetings: 45 },
      { id: 'sarah@example.com', type: 'indirect', meetings: 12 }
    ],
    edges: [
      { source: 'user@example.com', target: 'john@example.com', meetings: 45 },
      { source: 'john@example.com', target: 'sarah@example.com', meetings: 12 }
    ]
  },
  actionType: 'VIEW_GRAPH'
}

AI Helper Tools

summarize_text

Summarize provided text using AI.

Parameters:

Styles:

• concise - Brief, to the point
• detailed - Comprehensive summary
• bullet_points - Bullet point format

Response:

{
  success: true,
  summary: 'The meeting covered three main topics...',
  originalLength: 5432,
  summaryLength: 187
}

extract_info

Extract structured information from text.

Parameters:

Extraction Types:

• action_items - Tasks and action items
• decisions - Decisions made
• questions - Questions raised
• dates - Dates and deadlines
• people - People mentioned
• custom - Use custom schema

Response:

{
  success: true,
  extractionType: 'action_items',
  extracted: [
    {
      task: 'Send proposal to client',
      assignee: 'John',
      deadline: '2026-01-15',
      priority: 'high'
    }
  ]
}

answer_question

Answer a question about provided text.

Parameters:

Response:

{
  success: true,
  question: 'What was decided about the budget?',
  answer: 'The team approved a $50,000 budget for Q1...',
  confidence: 0.92,
  relevantExcerpts: [
    '...after discussion, we agreed to allocate $50,000...'
  ]
}

generate_response

Generate a formatted response from data.

Parameters:

Formats:

• markdown - Markdown formatted
• plain - Plain text
• html - HTML formatted

Response:

{
  success: true,
  response: '## Meeting Summary\n\n- **Date:** Jan 10, 2026\n...',
  format: 'markdown'
}

Action Tools

start_meeting

Start an instant meeting or join existing room.

Parameters:

Response:

{
  success: true,
  room: {
    name: 'instant-meeting-abc123',
    displayName: 'Quick Meeting',
    joinUrl: 'https://app.example.com/room/instant-meeting-abc123'
  },
  actionType: 'NAVIGATE_TO_ROOM'
}

create_room

Create a persistent room.

Parameters:

Settings:

{
  autoRecord: boolean,
  autoTranscribe: boolean,
  maxParticipants: number,
  waitingRoom: boolean,
  allowGuests: boolean
}

Response:

{
  success: true,
  room: {
    _id: 'stored_room_id',
    name: 'team-standup',
    displayName: 'Team Standup',
    settings: { ... },
    joinUrl: 'https://app.example.com/room/team-standup'
  },
  actionType: 'CREATE_ROOM'
}

create_schedule

Schedule a future meeting.

Parameters:

Recurring Settings:

{
  frequency: 'daily' | 'weekly' | 'monthly',
  interval: number,
  daysOfWeek: number[],  // 0=Sunday
  endDate: string,
  count: number
}

Response:

{
  success: true,
  schedule: {
    _id: 'schedule_id',
    title: 'Project Review',
    scheduledStart: '2026-01-15T14:00:00Z',
    scheduledEnd: '2026-01-15T15:00:00Z',
    roomName: 'project-review-abc123',
    attendees: ['attendee@example.com'],
    joinUrl: 'https://app.example.com/room/project-review-abc123'
  },
  actionType: 'SCHEDULE_CREATED'
}

cancel_schedule

Cancel a scheduled meeting.

Parameters:

Response:

{
  success: true,
  schedule: {
    _id: 'schedule_id',
    title: 'Project Review',
    status: 'cancelled',
    cancelledAt: '2026-01-11T10:00:00Z',
    cancelledBy: 'user@example.com',
    reason: 'Rescheduling to next week'
  },
  actionType: 'SCHEDULE_CANCELLED'
}

send_invites

Send meeting invitations via email.

Parameters:

*Either scheduleId or roomName is required.

Response:

{
  success: true,
  invitesSent: 3,
  recipients: ['user1@example.com', 'user2@example.com', 'user3@example.com'],
  failedRecipients: [],
  actionType: 'SEND_INVITES'
}

request_input

Request additional input from the user via form.

Parameters:

Field Definition:

{
  name: string,          // Field identifier
  label: string,         // Display label
  type: 'text' | 'email' | 'date' | 'select' | 'multiselect',
  required: boolean,
  options: string[],     // For select/multiselect
  defaultValue: any
}

Response:

{
  success: true,
  message: 'Please provide the following information:',
  fields: [
    { name: 'attendees', label: 'Attendees', type: 'multiselect' }
  ],
  actionType: 'NEEDS_INFO'
}

Related Documentation

• Getting Started - Quick start guide
• Hosted Server Guide - Using the hosted MCP servers
• Authentication - OAuth 2.1, MCP tokens, and API keys
• HTTP Transport - Streamable HTTP and SSE endpoints
• Architecture - System design and data flows