Skip to main content

​
Overview

Tavily MCP Server is available on Databricks Marketplace, enabling one-click installation that creates a secure Unity Catalog connection for authenticated access. Once installed, Tavily MCP can be used programmatically in your agent code (LangGraph, OpenAI, etc.) through Databricks-managed proxy endpoints, grounding your AI agents with real-time web search, extraction, crawling, and mapping.

​
Prerequisites

​
Setup

​
Install from Databricks Marketplace

1

Navigate to Marketplace and find Tavily MCP Server

In your Databricks workspace, go to Marketplace and search for Tavily MCP Server.
Search for Tavily MCP Server in Databricks Marketplace
2

Install and configure the connection

Click Install to install the Tavily MCP Server. In the installation dialog, configure the connection settings:
  • Connection name: Enter a name for the Unity Catalog connection (for example, tavily_mcp_connection).
  • Host: Pre-populated for Tavily.
  • Base path: Pre-populated for Tavily.
  • Bearer token: Enter your Tavily API Key as the bearer token.
Configure the connection for Tavily MCP Server
3

Verify Unity Catalog Connection

Navigate to Unity Catalog > Connect > Connections and verify that the connection has been created successfully.
Verify the connection from Unity Catalog

​
Share the MCP server connection

1

Open the Unity Catalog connection

Navigate to Unity Catalog > Connect > Connections and click on the connection you created earlier (for example, tavily_mcp_connection).
Open the Unity Catalog Connections
2

Grant access to the connection

Grant USE CONNECTION privileges to identity principals that need to use the Tavily MCP server connection. In your workspace, go to Permissions tab and grant USE CONNECTION to the identity principals that need to use the Tavily MCP server connection.
Grant USE CONNECTION privileges to identity principals

​
Test Tavily MCP Server within Databricks

You can test the Tavily MCP server directly within Databricks without writing any code. Using AI Playground:
1

Open AI Playground

Go to AI Playground in your Databricks workspace and choose a model with the Tools enabled label.
Select a model in AI Playground
2

Add Tavily MCP Server as a tool

Click Tools tab and select + Add tool and select MCP Servers from the available tool options. In the MCP Servers section, select External MCP servers to browse available connections, and choose the Unity Catalog connection you installed earlier (for example, tavily_mcp_connection).
Add Tavily MCP Server as a tool
3

Chat and test Tavily MCP Server

Chat with the LLM to test how it interacts with Tavily MCP tools.
Test Tavily MCP Server results

​
Add Tavily MCP Server to Databricks Assistant

1

Open Databricks Assistant

Go to Databricks Assistant in your Databricks workspace and click on the Settings icon.
2

Add MCP Server from Settings

In MCP Servers section, select + Add MCP Server. Go to External MCP servers dropdown and choose the Unity Catalog connection you installed earlier (for example, tavily_mcp_connection).
Add MCP Server from Settings

​
Use Tavily MCP in Your Agent Code

After installation, use Tavily MCP programmatically in your agent code by connecting to the proxy URL. The Databricks proxy makes external servers behave like managed MCP servers, handling authentication and token management.
1

Configure the proxy endpoint

Add the Tavily MCP proxy endpoint to your MANAGED_MCP_SERVER_URLS list. External MCP servers are proxied as managed servers, allowing you to use the same API for both:
from databricks.sdk import WorkspaceClient
from databricks_mcp import DatabricksMCPClient

# Initialize workspace client
workspace_client = WorkspaceClient()
host = workspace_client.config.host

# External MCP servers are proxied as managed servers, allowing you
# to use the same API for both managed and external servers
MANAGED_MCP_SERVER_URLS = [
    f"{host}/api/2.0/mcp/functions/system/ai",  # Default managed MCP
    f"{host}/api/2.0/mcp/external/tavily_mcp_connection"  # Tavily MCP proxy
]
2

Use with agents

Pass the proxy URL to the managed_server_urls parameter to create tools from both managed and external (proxied) servers:
# Use with agents — external servers work just like managed ones
import asyncio
from your_agent_code import create_mcp_tools  # Your agent's tool creation function

# Create tools from both managed and external (proxied) servers
mcp_tools = asyncio.run(
    create_mcp_tools(
        ws=workspace_client,
        managed_server_urls=MANAGED_MCP_SERVER_URLS
    )
)
3

Call tools directly (optional)

You can also call Tavily tools directly using the Databricks MCP Client:
# Direct tool call using DatabricksMCPClient
mcp_client = DatabricksMCPClient(
    server_url=f"{host}/api/2.0/mcp/external/tavily_mcp_connection",
    workspace_client=workspace_client
)

# List available tools
tools = mcp_client.list_tools()
print(f"Available tools: {[tool.name for tool in tools]}")

# Call a tool
response = mcp_client.call_tool(
    "tavily_search",
    {"query": "latest AI research breakthroughs"}
)
print(response.content[0].text)

​
Resources