Use Kubera with AI Assistants: MCP Setup Guide Part 2

Part 1 - ChatGPT, Claude, Grok, Perplexity

Jump to

Grok Bot

1. Open a chat with your Grok Bot agent.

2. Send the following

Add the Kubera connector.
MCP URL: ‘https://api.kubera.com/api/v1/mcp’
Client ID: ‘kubera-grok-bot’

3. Confirm when the agent asks to add it.

4. When the connect card appears, authorize your Kubera account.

5. After it’s connected, try asking: “What’s my net worth?” or “What’s my default portfolio?”

Cursor Desktop

1. Open a chat with your Cursor agent.

2. Send the following

Add the Kubera connector.
MCP URL: ‘https://api.kubera.com/api/v1/mcp’
Client ID: ‘kubera-cursor’

3. Confirm when the agent asks to add it.

4. When the connect card appears, authorize your Kubera account.

5. After it’s connected, try asking: “What’s my net worth?” or “What’s my default portfolio?”

Antigravity CLI

Step 1: Generate AUTH_TOKEN in Kubera

Kubera (on desktop) > Settings > AI Assistants > Create MCP Auth Token

Step 2: In Gemini CLI

ADD
agy mcp add --header "Authorization: Basic <AUTH_TOKEN>" kubera https://api.kubera.com/api/v2/mcp






REMOVE
agy mcp remove kubera 

Note: Kubera is not yet available on the Gemini consumer web/app versions. Export portfolio as LLM friendly markdown file, drop in Gemini and ask questions. Details below.

ChatGPT App

Step 1: Generate AUTH_TOKEN in Kubera

Kubera (on desktop) > Settings > AI Assistants > Create MCP Auth Token

Step 2: In Codex

Field Value
MCP Server Name kubera
Transport Type Streamable HTTP
URL https://api.kubera.com/api/v2/mcp
Header key Authorization
Header value Basic AUTH_TOKEN (see Step 1 above)
  • Replace the entire <AUTH_TOKEN>       placeholder with the token you generated in Kubera desktop:
  • See step 1 above to know how to generate AUTH_TOKEN
  • Keep the word Basic in front of the token.
  • For example, if your token is abc123yourtokenhere      , the Authorization header value should be: Basic abc123yourtokenhere                       
  • Click Save.

Others like Hermes, OpenClaw

Use this when your AI assistant is not listed above. Generate an MCP Auth Token in Kubera, then add a local MCP config.

Step 1: Generate AUTH_TOKEN in Kubera

Kubera (on desktop) > Settings > AI Assistants > Create MCP Auth Token

Step 2: Add the Kubera MCP server

In your agent’s MCP config (often mcp.json), add:

{
  "mcpServers": {
    "kubera": {
      "command": "npx",
      "args": [
        "mcp-remote",
        "https://api.kubera.com/api/v2/mcp",
        "--header",
        "Authorization: Basic ${AUTH_TOKEN}"
      ],
      "env": {
        "AUTH_TOKEN": "<paste-your-AUTH_TOKEN-here>"
      }
    }
  }
}

Replace <paste-your-AUTH_TOKEN-here> with the token from Step 1.

Keep the word Basic in the header. Paste only the token value into AUTH_TOKEN (do not put Basic inside the env value).

Restart or reload the agent so it picks up the new server.

Step 3: Start a new chat

Ask What’s my net worth?

Part 1 - ChatGPT, Claude, Grok, Perplexity

Related Articles