omi-mcp

Extended MCP tools for Omi REST API (https://api.omi.me).

Why This MCP?

Omi already provides an official MCP server for AI assistants. This project provides an alternative with additional capabilities:

What This MCP Adds

• Full Conversation Management: Create, update, delete conversations
• Action Items: Complete task/todo management (CRUD)
• Folders: List and organize folders
• API Key Management: Create and revoke API keys programmatically
• STDIO Transport: Works with Claude Desktop, Cursor, and all MCP clients

Features

This MCP server wraps the Omi REST API endpoints:

• Memories (5 tools): list, create, create-batch, update, delete
• Conversations (6 tools): list, create, create_from_segments, get, update, delete
• Folders (1 tool): list
• Action Items (5 tools): list, create, create-batch, update, delete
• API Keys (3 tools): list, create, revoke

Installation

Using uvx (recommended)

uvx omi-mcp

Using pip

pip install omi-mcp

Development

pip install -e ".[dev]"

Configuration

Authentication

Set one of these environment variables before starting the MCP server:

OMI_AUTHORIZATION_HEADER="Authorization: Bearer omi_mcp_XXXXX"
# or
AUTHORIZATION="Bearer omi_mcp_XXXXX"
# or
OMI_API_KEY="omi_mcp_XXXXX"

Usage

Running the MCP Server

# Run with stdio transport (for MCP clients)
python -m omi_mcp.server

Claude Desktop MCP Config

Add to your Claude MCP settings (claude_mcp.json):

{
  "mcpServers": {
    "omi-mcp": {
      "command": "uvx",
      "args": ["omi-mcp"]
    }
  }
}

Note: configure_api_key is removed. Configure auth via environment variables only. If you previously used configure_api_key(api_key="..."), set OMI_API_KEY="..." before starting the server.

Available Tools

Memories

• get_memories - List all memories
• create_memory - Create a new memory
• create_memories_batch - Create multiple memories
• update_memory - Update a memory
• delete_memory - Delete a memory

Conversations

• get_conversations - List all conversations
• create_conversation - Create a conversation
• create_conversation_from_segments - Create from transcript segments
• get_conversation - Get a conversation by ID
• update_conversation - Update a conversation
• delete_conversation - Delete a conversation

Folders

• get_folders - List all folders

Action Items

• get_action_items - List all action items
• create_action_item - Create a new action item
• create_action_items_batch - Create multiple action items
• update_action_item - Update an action item
• delete_action_item - Delete an action item

API Keys

• get_api_keys - List all API keys
• create_api_key - Create a new API key
• revoke_api_key - Revoke an API key

Development

# Install dev dependencies
pip install -e ".[dev]"

# Run tests
pytest

Releases

Publishing is automated through GitHub Actions.

• Push a version tag like v0.1.9
• The workflow creates a GitHub Release with built artifacts
• The same workflow publishes the package to PyPI