Overview
What is Vertica Mcp?
A Model Context Protocol (MCP) server for readonly communication with Vertica databases. It enables AI assistants like Claude and Cursor to safely execute SELECT, SHOW, DESCRIBE, EXPLAIN, and WITH queries, explore schemas, and retrieve metadata from Vertica.
How to use Vertica Mcp?
Configure an environment file with Vertica connection details (host, port, database, user, password), then add the server to your Cursor or Claude Desktop MCP configuration. Run via npx @hechtcarmel/[email protected] --env-file /path/to/vertica.env. The server exposes six tools for querying and schema discovery.
Key features of Vertica Mcp
- Readonly execution enforced on all queries
- Schema discovery (tables, views, indexes/projections)
- Table structure analysis with columns and constraints
- Streaming support for large result sets
- Parameterized queries with SQL injection protection
- SSL/TLS connection support
Use cases of Vertica Mcp
- AI assistant safely querying Vertica for analytics
- Exploring database schema and table definitions
- Debugging queries with EXPLAIN plans
- Streaming large fact tables into AI tools
- Integration with Cursor or Claude for Vertica data access
FAQ from Vertica Mcp
Is the server truly readonly?
Yes, it validates every query starts with SELECT, SHOW, DESCRIBE, EXPLAIN, or WITH. Write operations are blocked.
What runtime does it require?
Node.js 18.x or higher. It runs via npx and uses the official @modelcontextprotocol/sdk.
How do I configure the database connection?
Set environment variables: VERTICA_HOST, VERTICA_PORT, VERTICA_DATABASE, VERTICA_USER, VERTICA_PASSWORD. Optional: VERTICA_SSL, VERTICA_CONNECTION_LIMIT, VERTICA_QUERY_TIMEOUT.
What tools are available?
Six tools: execute_query, stream_query, get_table_structure, list_tables, list_views, list_indexes. Each accepts optional schema parameters.
How is performance handled?
Streaming queries support batching (default 1000 rows, max 10000) and max rows up to 1,000,000. Query timeouts prevent long-running operations.
