contextDB MCP Server

A Model Context Protocol (MCP) server that enables AI assistants to intelligently explore and query multiple databases simultaneously.

Features

• Multi-Database Support: Connect to Redshift, PostgreSQL, Aurora, and MySQL databases
• Metadata Discovery: Explore schemas, tables, columns, and relationships
• Data Profiling: Analyze table profiles, data quality, and distribution
• Query Execution: Execute SELECT queries with user approval
• Business Metrics: Generate 30-day business intelligence dashboards (Redshift)
• Performance Analysis: Analyze query performance and identify slow queries (Redshift)
• Cross-Database Search: Search for tables and columns across multiple datasources

Installation

# 1. Clone the repository
git clone https://github.com/antstackio/contextDB-mcp-server.git

# 2. Install with uv tool from the root directory
uv tool install .

# 3. Add uv's bin directory to PATH (one-time setup)
echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.zshrc
source ~/.zshrc

This makes the contextDB command available globally.

Configuration

1. Create Datasources Configuration

Copy the example configuration and update with your database credentials:

cp datasources.example.json datasources.json

Edit datasources.json:

2. Configure MCP Client

Claude Code

Create .settings.json in your project root:

{
  "mcpServers": {
    "contextDB-mcp": {
      "command": "contextDB",
      "env": {
        "DATASOURCES_CONFIG": "/path/to/your/datasources.json"
      }
    }
  }
}

Gemini CLI

Create .gemini/settings.json in your project root:

{
  "mcpServers": {
    "contextDB-mcp": {
      "command": "contextDB",
      "env": {
        "DATASOURCES_CONFIG": "/path/to/your/datasources.json"
      }
    }
  }
}

Cursor

Open Command Palette and select "Add new global MCP server". This opens the MCP settings file where you can add:

{
  "mcpServers": {
    "contextDB-mcp": {
      "command": "contextDB",
      "env": {
        "DATASOURCES_CONFIG": "/path/to/your/datasources.json"
      }
    }
  }
}

Usage

Discovery Tools

1 . Get datasource overview - Start with this to get started

get_datasource_overview_tool(datasource_id="analytics_warehouse")

2. List all datasources

list_all_datasources_tool()

3. Search across all datasources

search_across_datasources_tool(keyword="customer")

Database Exploration Tools

4. Get database overview

get_database_overview_tool(datasource_id="analytics_warehouse")

5. List schemas

list_schemas_tool(datasource_id="analytics_warehouse")

6. List tables in a schema

list_tables_tool(datasource_id="analytics_warehouse", schema_name="public")

7. Discover schema metadata

discover_schema_metadata_tool(
    datasource_id="analytics_warehouse",
    schema_name="public"
)

8. Get table profile

get_table_profile_tool(
    datasource_id="analytics_warehouse",
    schema_name="public",
    table_name="customers"
)

Query and Analysis Tools

9. Execute SQL query (requires approval)

execute_sql_tool(
    datasource_id="analytics_warehouse",
    sql_query="SELECT * FROM public.customers LIMIT 10"
)

10. Check data quality

check_data_quality_tool(
    datasource_id="analytics_warehouse",
    schema_name="public",
    table_name="customers"
)

Redshift-Specific Tools

11. Analyze query performance

analyze_query_performance_tool(
    datasource_id="analytics_warehouse",
    limit=10
)

12. Get business metrics

get_business_metrics_tool(
    datasource_id="analytics_warehouse",
    schema_name="public",
    days=30
)

Environment Variables

• DATASOURCES_CONFIG: Path to datasources configuration file (default: datasources.json)
• TRANSPORT: Set to http for HTTP mode - can be used to host the server (default: stdio)

Running in HTTP Mode

For hosting or testing:

# Using uv
uv run contextDB --http

# Or set environment variable
TRANSPORT=http uv run contextDB

The server will start on http://127.0.0.1:8000

Supported Database Types

• Redshift: Full feature support including performance analysis and business metrics
• PostgreSQL: Core features (schemas, tables, queries, data quality)
• Aurora PostgreSQL: Core features
• MySQL: Core features

Security Notes

• All SQL queries are READ-ONLY (SELECT statements only)
• User approval required for query execution
• Credentials stored in local configuration file

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

MIT

Requirements

• Python >=3.10
• Database drivers (installed automatically):
  • asyncpg for PostgreSQL/Redshift/Aurora
  • redshift-connector for Redshift-specific features
• MCP SDK

Troubleshooting

Connection Issues

1. Verify your datasources.json credentials
2. Check network connectivity to database
3. Ensure database allows connections from your IP
4. Verify database user has necessary permissions