Connect to and Query Vertica Databases
MCP server for Vertica exposing query, streaming, bulk-load, and schema-inspection tools with fine-grained write permissions.
Why it matters
Establish secure connections to Vertica databases, execute complex SQL queries, and manage data operations with fine-grained access control.
Outcomes
What it gets done
Manage database connections with pooling and SSL support.
Execute SQL queries and stream large result sets.
Perform bulk data loading using the COPY command.
Inspect table schemas, indexes, and views.
Install
Add it to your toolbox
Run in your project directory:
curl -fsSL https://spark.entire.vc/get/vb-vertica | bash Capabilities
Tools your agent gets
Execute SQL queries with support for all SQL operations
Stream large query results in batches with configurable batch size
Bulk load data using the COPY command, efficient for large datasets
Retrieve detailed table structure, including column information and constraints
List all indexes for a table with index type, uniqueness, and column information
List all views in a schema with view definitions
Overview
Vertica MCP Server
This MCP server connects to Vertica, exposing query execution/streaming, COPY-based bulk loading, and schema inspection (table structure, indexes, views) as tools, with layered connection, operation, and per-schema permissions. Use it when an agent needs to query, stream, bulk-load, or inspect a Vertica database's schema - write and DDL operations are disabled by default and must be explicitly enabled per operation and schema.
What it does
Provides an MCP server for Vertica - described as the first implementation of a Vertica MCP server, listed in the official Model Context Protocol Registry - exposing database connection management, query operations, and schema inspection as MCP tools.
Connection management includes connection pooling with configurable limits, SSL/TLS support, automatic connection cleanup, and connection timeout handling. Six tools are exposed: execute_query (runs SQL queries, supporting all SQL operations), stream_query (streams large query results in configurable batches), copy_data (bulk data loading via Vertica's COPY command, efficient for large datasets), get_table_structure (detailed table structure, column information, constraints), list_indexes (all indexes for a table, with index type/uniqueness/column info), and list_views (all views in a schema, with view definitions). Security features layer operation-level permissions (INSERT, UPDATE, DELETE, DDL, each independently toggleable) with schema-specific permission overrides, plus SSL/TLS support and password masking in logs.
{
"mcpServers": {
"vertica": {
"command": "uvx",
"args": ["mcp-vertica"],
"env": {
"VERTICA_HOST": "localhost",
"VERTICA_PORT": 5433,
"VERTICA_DATABASE": "VMart",
"VERTICA_USER": "dbadmin",
"VERTICA_PASSWORD": "test_password",
"VERTICA_CONNECTION_LIMIT": 10,
"VERTICA_SSL": false,
"VERTICA_SSL_REJECT_UNAUTHORIZED": true
}
}
}
}
Configuration is environment-variable driven: connection settings (VERTICA_HOST, VERTICA_PORT, VERTICA_DATABASE, VERTICA_USER, VERTICA_PASSWORD, VERTICA_CONNECTION_LIMIT, VERTICA_SSL, VERTICA_SSL_REJECT_UNAUTHORIZED), operation permissions (ALLOW_INSERT_OPERATION, ALLOW_UPDATE_OPERATION, ALLOW_DELETE_OPERATION, ALLOW_DDL_OPERATION, all boolean), and per-schema permissions (SCHEMA_INSERT_PERMISSIONS, SCHEMA_UPDATE_PERMISSIONS, SCHEMA_DELETE_PERMISSIONS, SCHEMA_DDL_PERMISSIONS, each a comma-separated schema:bool list). It can also run via Docker or uvx with CLI args instead of environment variables. Debug logging is available in Docker via a DEBUG env var (0=none through 3=maximum verbosity, -vvv) and an EXTRA_ARGS variable for passing additional CLI flags.
When to use - and when NOT to
Use it when an AI agent needs to query, stream, bulk-load, or inspect the schema of a Vertica database directly, with fine-grained control over which write/DDL operations are allowed per schema. Write and DDL operations are disabled by default (ALLOW_*_OPERATION=false) and must be explicitly enabled - review the permission configuration before allowing any agent to perform inserts, updates, deletes, or DDL against production data.
Inputs and outputs
Configuration is Vertica connection details (host, port, database, user, password, connection limit, SSL settings) plus operation and schema-level permission flags, supplied via environment variables or CLI args. Tool inputs are SQL queries, table/schema names, or COPY-command data; outputs are query results (direct or streamed in batches), table/index/view metadata, or load-operation results.
Capabilities
For local development and testing, the project ships a Docker Compose example that spins up a Vertica Community Edition container (vertica/vertica-ce) bound to ports 5433 and 5444, with a persisted VMart data volume and a vsql-based healthcheck - letting the server be exercised against a real Vertica instance without a production connection. The project is MIT licensed.
How to install
Install automatically via Smithery (npx -y @smithery/cli install @nolleh/mcp-vertica --client claude), or manually by configuring your MCP client with uvx mcp-vertica (or the Docker image nolleh/mcp-vertica) and the environment variables shown above. Available as a PyPI package (mcp-vertica).
Who it's for
Teams running Vertica who want an AI agent to query, stream, bulk-load, or inspect their database schema directly, with explicit, per-schema control over which write and DDL operations the agent is permitted to run.
Source README
MCP Vertica
š First implementation of Vertica MCP Server ⢠Learn more
ā Listed in Model Context Protocol Official Registry
A Vertica MCP(model-context-protocol) Server
Example: MCP Server Setting
Create or edit the file your mcp client config file with the following content:
UVX
{
"mcpServers": {
"vertica": {
"command": "uvx",
"args": ["mcp-vertica"],
"env": {
"VERTICA_HOST": "localhost",
"VERTICA_PORT": 5433,
"VERTICA_DATABASE": "VMart",
"VERTICA_USER": "dbadmin",
"VERTICA_PASSWORD": "test_password",
"VERTICA_CONNECTION_LIMIT": 10,
"VERTICA_SSL": false,
"VERTICA_SSL_REJECT_UNAUTHORIZED": true
}
}
}
}
Or with args
{
"mcpServers": {
"vertica": {
"command": "uvx",
"args": [
"mcp-vertica",
"--host=localhost",
"--db-port=5433",
"--database=VMart",
"--user=dbadmin",
"--password=test_password",
"--connection-limit=10"
]
}
}
}
Docker
{
"mcpServers": {
"vertica": {
"command": "docker",
"args": ["run", "-i", "--rm", "nolleh/mcp-vertica"],
"env": {
"VERTICA_HOST": "localhost",
"VERTICA_PORT": 5433,
"VERTICA_DATABASE": "VMart",
"VERTICA_USER": "dbadmin",
"VERTICA_PASSWORD": "test_password",
"VERTICA_CONNECTION_LIMIT": 10,
"VERTICA_SSL": false,
"VERTICA_SSL_REJECT_UNAUTHORIZED": true
}
}
}
}
Features
Database Connection Management
- Connection pooling with configurable limits
- SSL/TLS support
- Automatic connection cleanup
- Connection timeout handling
Query Operations
- Execute SQL queries
- Stream large query results in batches
- Copy data operations
- Transaction management
Schema Management
- Table structure inspection
- Index management
- View management
- Constraint information
- Column details
Security Features
- Operation-level permissions (INSERT, UPDATE, DELETE, DDL)
- Schema-specific permissions
- SSL/TLS support
- Password masking in logs
Tools
Database Operations
execute_query- Execute SQL queries
- Support for all SQL operations
stream_query- Stream large query results in batches
- Configurable batch size
copy_data- Bulk data loading using COPY command
- Efficient for large datasets
Schema Management
get_table_structure- Get detailed table structure
- Column information
- Constraints
list_indexes- List all indexes for a table
- Index type and uniqueness
- Column information
list_views- List all views in a schema
- View definitions
Configuration
Environment Variables
VERTICA_HOST=localhost
VERTICA_PORT=5433
VERTICA_DATABASE=VMart
VERTICA_USER=newdbadmin
VERTICA_PASSWORD=vertica
VERTICA_CONNECTION_LIMIT=10
VERTICA_SSL=false
VERTICA_SSL_REJECT_UNAUTHORIZED=true
Operation Permissions
ALLOW_INSERT_OPERATION=false
ALLOW_UPDATE_OPERATION=false
ALLOW_DELETE_OPERATION=false
ALLOW_DDL_OPERATION=false
Schema Permissions
SCHEMA_INSERT_PERMISSIONS=schema1:true,schema2:false
SCHEMA_UPDATE_PERMISSIONS=schema1:true,schema2:false
SCHEMA_DELETE_PERMISSIONS=schema1:true,schema2:false
SCHEMA_DDL_PERMISSIONS=schema1:true,schema2:false
Installation
Installing via Smithery
To install Vertica Database Connector for Claude Desktop automatically via Smithery:
npx -y @smithery/cli install @nolleh/mcp-vertica --client claude
Installing Manually
Open your favorite mcp client's config file, then configure with uvx mcp-vertica
Development
Debug Mode
When running with Docker, you can enable debug logging by setting the DEBUG environment variable:
# Run with maximum verbosity (-vvv)
docker run -e DEBUG=3 -e VERTICA_HOST=localhost ... nolleh/mcp-vertica:latest
# Run with medium verbosity (-vv)
docker run -e DEBUG=2 -e VERTICA_HOST=localhost ... nolleh/mcp-vertica:latest
# Pass additional arguments
docker run -e EXTRA_ARGS="--connection-limit=20" -e VERTICA_HOST=localhost ... nolleh/mcp-vertica:latest
In docker-compose.yml:
environment:
DEBUG: 3 # 0=none, 1=-v, 2=-vv, 3=-vvv
EXTRA_ARGS: "--connection-limit=20" # Optional additional arguments
Appendix: For Testing, VerticaDB Docker Compose Example
version: "3.8"
services:
vertica:
# image: vertica/vertica-ce:11.1.0-0
image: vertica/vertica-ce:latest
platform: linux/amd64
container_name: vertica-ce
environment:
VERTICA_MEMDEBUG: 2
ports:
- "5433:5433"
- "5444:5444"
volumes:
- vertica_data:/home/dbadmin/VMart
healthcheck:
test:
[
"CMD",
"/opt/vertica/bin/vsql",
"-h",
"localhost",
"-d",
"VMart",
"-U",
"dbadmin",
"-c",
"SELECT 1",
]
interval: 10s
timeout: 5s
retries: 5
start_period: 30s
restart: unless-stopped
mcp-vertica:
image: nolleh/mcp-vertica:latest
container_name: mcp-vertica
ports:
- "8081:8081"
environment:
# Transport mode
TRANSPORT: http
PORT: 8081
# Debug settings (0=none, 1=-v, 2=-vv, 3=-vvv)
DEBUG: 3 # Set to 3 for maximum verbosity
# Extra command line arguments (optional)
# EXTRA_ARGS: "--some-flag"
# Vertica connection settings
VERTICA_HOST: vertica
VERTICA_PORT: 5433
VERTICA_DATABASE: VMart
VERTICA_USER: dbadmin
VERTICA_PASSWORD: ""
VERTICA_CONNECTION_LIMIT: 10
VERTICA_SSL: "false"
depends_on:
vertica:
condition: service_healthy
volumes:
vertica_data:
driver: local
Then run server by following instruction Example: Mcp Server Setting,
Then see everything works as fine
FAQ
Common questions
Discussion
Questions & comments Ā· 0
Sign In Sign in to leave a comment.
