Vertica MCP Server

A Model Context Protocol (MCP) server for Vertica databases. Enables AI assistants to query and explore Vertica databases through natural language.

Safety-first design: Readonly mode by default. Write operations require explicit configuration.

Features

• 6 MCP Tools: Query execution, streaming, schema discovery

• Readonly Protection: Only SELECT/SHOW/DESCRIBE/EXPLAIN/WITH queries by default

• Large Dataset Streaming: Efficient batch processing (up to 1M rows)

• Vertica-Optimized: Projection awareness, columnar query support

• Production Ready: Connection pooling, SSL support, timeout configuration

• Parameter Binding: SQL injection protection

Quick Start

Claude Code

Create your .env file with connection details:

Cursor

1. Create environment file ~/.cursor/vertica.env:

2. Configure ~/.cursor/mcp.json:

3. Restart Cursor

Claude Desktop

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json Windows: %APPDATA%/Claude/claude_desktop_config.json

Configuration

Required Variables

Optional Variables

Enabling Write Operations

To allow INSERT/UPDATE/DELETE/CREATE/DROP operations:

Warning: Only disable readonly mode if you understand the implications.

Available Tools

Query Execution

• execute_query: Execute SQL with optional parameters

• stream_query: Handle large datasets with configurable batching

Schema Discovery

• get_table_structure: Table columns, types, constraints

• list_tables: All tables in schema with metadata

• list_views: All views with definitions

• list_indexes: Vertica projections for optimization

Usage Examples

Query Data

Explore Schema

Analyze Performance

Stream Large Results

When querying large datasets, use the stream_query tool:

• Default batch size: 1000 rows

• Configurable batch size: 1-10,000 rows

• Maximum rows: 1,000,000

Troubleshooting

Connection Failed

Verify:

• Host and port are reachable

• Database credentials are correct

• User has required permissions

Permission Errors

• User needs SELECT permissions on tables

• User needs access to system catalogs (v_catalog.*)

Query Timeouts

Increase timeout for complex queries:

Large Result Sets

Use stream_query instead of execute_query for queries returning >10,000 rows.

Requirements

• Node.js >= 18.0.0

• Vertica database (any recent version)

• Network access to Vertica server

Support

• Issues: GitHub Issues

• Releases: GitHub Releases

License

MIT License - see LICENSE file.

Acknowledgments

This project's architecture and tool design are based on mcp-vertica by @nolleh.

Current Version: 1.3.5