Overview
Model Context Protocol (MCP) allows AI assistants like Claude to interact with Plumi directly. Query your databases, explore schemas, and analyze data through natural conversation.What is MCP?
MCP is an open protocol that enables AI assistants to:- Execute queries on your connected databases
- Explore database schemas
- Access query results and analytics
- Maintain context across conversations
Setting Up MCP
Step 1: Get Your MCP URL
Each Plumi workspace has a unique MCP endpoint:Step 2: Configure Your AI Assistant
Claude Desktop
Add to your Claude Desktop config (~/.config/claude/claude_desktop_config.json):
Cursor
Add to your Cursor MCP settings:Available Tools
Once connected, your AI assistant can use these tools:query
Execute SQL queries on your connected databases.
Example prompt: “Query the orders table to show total revenue by month”
list_tables
List all tables in a connected database.
Example prompt: “What tables are available in my PostgreSQL database?”
describe_table
Get schema details for a specific table.
Example prompt: “Describe the structure of the customers table”
list_connections
List all configured database connections.
Example prompt: “What databases do I have connected?”
Example Conversations
Data Exploration
You: What tables are in my analytics database? Claude: Let me check your database schema… Your analytics database contains these tables:
users(12 columns)orders(8 columns)products(6 columns)order_items(5 columns)
Running Queries
You: Show me the top 10 customers by total spend Claude: I’ll query your database for that information… Here are your top 10 customers by total spend:
Customer Total Spend Acme Corp $125,000 … …
Security
API Key Management
- Generate API keys in Settings > API Keys
- Use separate keys for different integrations
- Rotate keys regularly
- Revoke keys immediately if compromised
Permissions
MCP connections respect your database permissions:- Read-only database users can only query data
- No write operations are exposed through MCP
Troubleshooting
Connection Failed
- Verify your API key is valid
- Check your workspace URL is correct
- Ensure your Plumi workspace is running
No Tools Available
- Restart your AI assistant after config changes
- Verify the MCP config JSON is valid
- Check for typos in the URL