Skip to main content
Claude Code is the command line version of Anthropic’s AI assistant, Claude. Once successfully connected to Omni, you can interact with your Omni data right in your Claude Code session. Looking for Claude Desktop? Check out the Claude Desktop guide.

​
Requirements

Every person who wants to use the MCP Server must complete the setup.
To follow the steps in this guide, you’ll need:

​
Setup

1

Add the Omni MCP server

In your terminal, run the following command:
claude mcp add --transport http omni https://callbacks.omniapp.co/callback/mcprun
2

Start a new Claude session

Run claude to start a new session.
3

Authenticate with Omni

  1. Run /mcp to open the MCP servers list:
  2. Navigate to and select the Omni MCP server:
  3. Press Enter to select the Authenticate option.
  4. Your browser will open and redirect you to an Omni authorization page.
  5. Review the requested permissions and click Authorize.
Omni will automatically create a Personal access token and link it to your Omni user account.
4

Verify the connection

At this point, you should be able to use the Omni MCP server in your Claude Code sessions:

​
Troubleshooting

​
Tool denial in ‘don’t ask’ mode

Claude Code may sometimes return an error similar to: 
Permission to use mcp__omni__pickModel has been denied because Claude Code is running in don't ask mode
This mode automatically blocks tools that aren’t pre-approved rather than prompted you for approval. To resolve this issue, you have a few options:
  • Restart Claude Code with the tools allowed: 
    claude --allowed-tools 'mcp__omni__*'
  • Update your permission settings to allow MCP tools. Add mcp__omni__* to your allowed tools in your Claude Code settings or .claude/settings.json
  • Switch out of ‘don’t ask’ mode, which will ensure you are prompted to approve tool calls