Skip to main content

​
Prerequisites

  • A Claude account (Pro or Team plan required for integrations)
  • A VexiData account with at least one connected database

​
Connect Claude to VexiData

1

Copy your MCP Server URL

Your MCP Server URL is:
You can also find this in Settings > AI Integrations in your VexiData AI Integrations page.
2

Add VexiData as a connector in Claude

  1. Go to claude.ai/settings/integrations
  2. Click Add Custom Connector
  3. Paste your VexiData MCP Server URL
  4. Click Add and authorize access when prompted
3

Enable VexiData in your conversation

After connecting, make sure VexiData is enabled for your conversations:
  1. In the Claude chat window, click the settings button
  2. Ensure VexiData is enabled
Once enabled, you can ask questions about your databases directly in the conversation. For example:
  • “How many active users do we have?”
  • “Show me orders from the past 7 days”
  • “What are the most popular product categories?”
Claude will use VexiData to run the appropriate SQL query against your connected database and return the results.

​
Managing your connection

You can view and manage your Claude connection from Settings > AI Integrations in VexiData. From there you can:
  • See when the connection was established
  • Disconnect Claude by clicking Disconnect

​
What VexiData exposes to Claude

Claude connects to VexiData over the Model Context Protocol via OAuth 2.0. All operations are read-only — no data is written to your database.

​
Tools

​
Resources

​
Prompts

​
Example conversations

​
Example 1 — Explore what’s in a database

You: “What databases do I have connected and what’s in them?”
Claude calls get_user_details to find your connections, then list_tables on the one you pick. It returns a summary of tables with descriptions and row counts, and suggests areas worth exploring.

​
Example 2 — Answer a business question

You: “Show me the top 10 customers by total order value this year.”
Claude calls list_tables to locate the relevant tables, get_table_schema with table_names: ["customers", "orders"] to read the columns and relationships, then execute_query to run a JOIN with aggregation and a date filter. The result is a formatted table of the top 10 customers.

​
Example 3 — Export data as CSV

You: “Export all orders from last month as CSV.”
Claude calls get_table_schema to confirm the date column on the orders table, then export_query_as_csv with a SELECT … WHERE created_at … query. You receive a temporary download link valid for 5 minutes.

​
Troubleshooting

  • MCP integrations require a Claude Pro or Team plan. Free accounts do not have access to integrations
  • Make sure you’re on claude.ai, not the API console
  • Ensure you’re logged in to VexiData before approving the connection
  • Try copying the MCP Server URL again from VexiData and re-adding it in Claude
  • Clear your browser cookies for vexidata.com and try again
  • Verify you have at least one database connected in VexiData under Data Sources
  • Check that your database connection is active (the schema should be analyzed)
  • Try being specific about which database or table you’re asking about