DBCode

What the DBCode MCP server does

The DBCode MCP server connects AI agents to databases through connections already configured in DBCode. It provides database access through built-in MCP tools, allowing an agent to query data, inspect schema information, generate SQL queries, and create tables.

DBCode is also a database client for VS Code and compatible editor forks. The broader product supports more than 90 databases, warehouses, lakehouses, and file formats. Examples named in the project include PostgreSQL, MySQL, SQL Server, SQLite, Oracle, MongoDB, Redis, DuckDB, Snowflake, BigQuery, Databricks, ClickHouse, Cassandra, Elasticsearch, Neo4j, Firebase, DynamoDB, and MariaDB.

How it works

Agents use the connections that are already available in DBCode rather than defining a separate connection for each agent. Permissions can be set per connection, which lets a developer control the database access available to AI tools.

The MCP functionality is built into the DBCode editor extension. In VS Code, MCP tools are registered automatically for Copilot. Cursor also receives the tools automatically. Antigravity, Windsurf, and Kiro provide a one-click connection flow, while external agents such as Claude Code can connect through the DBCode: Connect an AI Agent command.

Setup and configuration

Install DBCode from the VS Code Marketplace or OpenVSX. OpenVSX is listed for Cursor, Windsurf, and other VS Code forks. Configure the required database connections in DBCode, then grant the appropriate AI permissions for each connection.

For Claude Code and other external agents, use the DBCode: Connect an AI Agent command from the editor. The material does not specify environment variables or a standalone package installation procedure, so setup depends on the DBCode extension and the client connection flow.

Tools and capabilities

The DBCode MCP server exposes the database operations described in the project documentation, including:

• Querying connected databases
• Asking questions about database schemas
• Generating SQL queries from natural-language requests
• Creating tables through an AI agent
• Using existing DBCode connections with per-connection permissions

The DBCode extension also includes SQL editing with database-specific autocomplete, table and column type suggestions, data filtering and editing, relationship navigation, ER diagrams, query result charts, SQL and Python notebooks, and execution-plan visualization. These are DBCode features; the provided material does not state that every feature is exposed as an MCP tool.

Limitations and notes

The server depends on connections configured in DBCode and on the permissions assigned to those connections. The README describes support for more than 90 database-related systems and formats, but it does not provide a complete list of MCP-specific operation coverage for each one.

The DBCode MCP server is documented as part of the DBCode editor extension rather than as a separately described hosted service. The provided material also does not specify transport details, environment variables, authentication settings, or a standalone command-line installation.