Dbeaver

bachden1starsClaim this server

DBeaver MCP Server is an Eclipse/OSGi plugin that runs inside DBeaver and exposes configured database connections to AI agents through a local MCP Streamable HTTP server.

Databases

Install Run on PlugMCPClaim this server

GitHub

Run on PlugMCP

1# Dbeaver

README.mdMCP server

Transport: stdio
Status: Local

Power your agent

Add this server

Install from the publisher. Not a PlugMCP endpoint.

Run on PlugMCP

Host a copy with our keys and logs.

No packaged release in the registry yet. Follow the repository README.

RepositoryPublisher installStep-by-step: add this in Claude

Remote URL or a command on your machine? How the two installs differ

Run on PlugMCP

Host a copy with our keys and logs.

A new server you create. It is not this listing's official endpoint. Use the publisher install when you want their URL and their credentials.

Launch

Server facts

Local

Transport: stdio
Last checked: 2026-10-28

Stars: 1
Tools: —
Uptime: —

Tools

No tool list yet. Remote servers get one after the first health check.

FAQ

Questions

How do I install Dbeaver?

Copy the stdio config for Cursor, Claude Desktop, Claude Code, or VS Code on this page. It uses the publisher's install, not a PlugMCP proxy.

Does Dbeaver need an API key?

No required environment variable is listed on this record.

What transport does Dbeaver use?

Dbeaver runs locally over stdio.

When was Dbeaver last checked?

Last checked 2026-10-28. DBeaver MCP Server is an Eclipse/OSGi plugin that runs inside DBeaver and exposes configured database connections to AI agents through a local MCP Streamable HTTP server.

From the publisher

Documentation

DBeaver MCP Server

DBeaver MCP Server is an Eclipse/OSGi plugin that runs inside DBeaver and exposes configured database connections to AI agents through a local MCP Streamable HTTP server.

The plugin uses DBeaver's connection registry, drivers, and live sessions directly. It does not read or decrypt credentials-config.json.

Install From GitHub Pages

In DBeaver, open Help -> Install New Software....

Add the following update site:

https://bachden.github.io/dbeaver-mcp-server/

Select DBeaver MCP Server and complete the installation wizard.
Restart DBeaver when prompted.

This URL is a p2 repository for DBeaver. The root may not render an HTML page in a browser; DBeaver reads p2.index, artifacts.xml.xz, and content.xml.xz directly.

Human In The Loop

The plugin follows a what you see is what you get rule:

list_connections lists both open and closed connections and includes a connected field.
get_schema, run_query, and execute_sql only use connections that are already open in DBeaver.
The plugin never opens a database connection automatically or presents a credential prompt on the user's behalf.
When a connection is closed, the tool returns an error and asks the user to open it manually in DBeaver first.
Tunnels, credentials, MFA, and the decision to connect remain under the user's control in the DBeaver UI.

Tools

Tool	Description
`list_connections`	Lists connections, their `connectionId`, endpoint, and `connected` state.
`get_schema`	Browses catalogs, schemas, tables, and columns through a dotted `path`.
`run_query`	Runs read-only SQL such as SELECT, WITH, SHOW, EXPLAIN, and DESCRIBE.
`execute_sql`	Runs any SQL, including INSERT, UPDATE, DELETE, and DDL. It can modify or delete data.

All three database-access tools reject closed connections.

list_connections accepts an optional name filter. The default nameMode: "ilike" supports SQL-style % and _ wildcards with case-insensitive matching. Set nameMode: "regex" to use a Java regular expression with substring matching.

run_query and execute_sql accept an optional showOnUi boolean (default false). When it is true, the statement executes once through a visible DBeaver SQL editor: its result appears in the DBeaver result grid and the same data is returned to the MCP caller. DBeaver's normal execution confirmations remain active, and the connection must already be open.

Server And UI

The default endpoint is:

POST http://127.0.0.1:8722/mcp

The DBeaver MCP menu provides:

Start Server
Stop Server
Connections...
Settings...

The status bar displays the current server state. The Start/Stop choice is persisted in the workspace: a stopped server remains stopped after DBeaver restarts, while an enabled server starts again automatically.

The bundle is activated through the org.jkiss.dbeaver.pluginService lifecycle after the DBeaver platform is ready. The p2 metadata does not force markStarted, which avoids Equinox trying to start a duplicate singleton module.

The Window -> Preferences -> DBeaver MCP Server page configures the listener host, port, bearer-token authentication, and client snippets for Claude Code, Claude Desktop, Codex CLI, or raw HTTP.

The server supports initialize, ping, tools/list, tools/call, and JSON-RPC batches. It does not provide an SSE stream.

Project Layout

dbeaver-mcp-server-plugin-parent
├── pom.xml
├── .github/workflows/
├── dbeaver-mcp-server-plugin
│   ├── META-INF/MANIFEST.MF
│   ├── plugin.xml
│   ├── dbeaver-ci.target
│   ├── pom.xml
│   └── src/
├── dbeaver-mcp-server-plugin.feature
│   ├── feature.xml
│   └── pom.xml
└── dbeaver-mcp-server-plugin.updatesite
    ├── category.xml
    └── pom.xml

The modules are:

dbeaver-mcp-server-plugin-parent: the Maven/Tycho reactor.
dbeaver-mcp-server-plugin: the OSGi bundle.
dbeaver-mcp-server-plugin.feature: the installable p2 feature.
dbeaver-mcp-server-plugin.updatesite: the p2 repository.

Development Setup

Requirements:

DBeaver 23+ installed.
Java 21.
Eclipse IDE with PDE and m2e.
Maven 3.9 or newer.

Import the repository root with File -> Import -> Existing Maven Projects. Eclipse will discover the four reactor projects.

Open dbeaver-mcp-server-plugin/dbeaver-ci.target, select Set as Active Target Platform, and wait for PDE to finish resolving it. This is the same public p2 target definition consumed by Tycho and GitHub Actions. No local DBeaver installation path is part of the build.

Build The p2 Update Site

Run from the repository root:

mvn clean verify

Tycho builds the bundle, feature, and p2 repository. The complete repository is copied to:

dist/
├── artifacts.jar
├── content.jar
├── features/
├── plugins/
└── p2.index

GitHub Actions runs the same mvn clean verify command and uses dbeaver-ci.target through the parent POM. dist/ is the stable local update-site location. Each build replaces the 0.1.6.qualifier qualifier with a timestamp so p2 can identify a new update.

Install From A Local Update Site

Build the project to create dist/.
In DBeaver, open Help -> Install New Software....
Select Add... -> Local... and choose the dist/ directory.
Give the repository a name such as Local DBeaver MCP.
Select DBeaver MCP Server and complete the wizard.
Restart DBeaver when prompted.

There is no need to copy JARs, edit bundles.info, or use dropins/. Normal installation and update flows do not require -clean.

Update The Plugin

For local development:

Rebuild the update site:
```
mvn clean verify
```
In DBeaver, select Help -> Check for Updates.
Install the new version and restart DBeaver when prompted.

The local repository remains at dist/, so it does not need to be added again after every build. Do not uninstall and reinstall the same timestamped version. Create a new qualifier or increment the base version, then use Check for Updates.

If the Error Log reports Another singleton bundle selected after the exact same version has been reinstalled, close DBeaver completely and run it once with:

/Applications/DBeaver.app/Contents/MacOS/dbeaver -clean

DBeaver can be started normally after that one clean launch.

GitHub Actions And Pages

The repository contains two workflows:

Build and Publish DBeaver Update Site runs on pushes to main, builds with the shared dbeaver-ci.target, and deploys dist/ to GitHub Pages.
Validate DBeaver Update Site runs for pull requests and uploads dist/ as a build artifact.

The published update site is:

https://bachden.github.io/dbeaver-mcp-server/

Do not commit target/ or dist/; Maven recreates these artifacts in CI.

Connect An AI Agent

Claude Code example with authentication disabled:

claude mcp add --transport http dbeaver http://127.0.0.1:8722/mcp

When authentication is enabled, clients must send:

Authorization: Bearer <token>

Codex CLI example:

[mcp_servers.dbeaver]
url = "http://127.0.0.1:8722/mcp"

[mcp_servers.dbeaver.http_headers]
Authorization = "Bearer <token>"

Security

The server binds to 127.0.0.1 by default.
Enable bearer-token authentication when binding to a LAN address or 0.0.0.0.
execute_sql accepts unrestricted SQL and can modify or delete data.
The run_query filter is not a replacement for database permissions. Use a read-only database account when a hard write restriction is required.
The user must open a connection in DBeaver before an agent can access that database.
The server can be stopped at any time from the menu or status bar.

Dbeaver

bachden1starsClaim this server

DBeaver MCP Server is an Eclipse/OSGi plugin that runs inside DBeaver and exposes configured database connections to AI agents through a local MCP Streamable HTTP server.

Databases

Install Run on PlugMCPClaim this server

GitHub

Run on PlugMCP

1# Dbeaver

README.mdMCP server

Transport: stdio
Status: Local

Power your agent

Add this server

Install from the publisher. Not a PlugMCP endpoint.

Run on PlugMCP

Host a copy with our keys and logs.

No packaged release in the registry yet. Follow the repository README.

RepositoryPublisher installStep-by-step: add this in Claude

Remote URL or a command on your machine? How the two installs differ

Run on PlugMCP

Host a copy with our keys and logs.

A new server you create. It is not this listing's official endpoint. Use the publisher install when you want their URL and their credentials.

Launch

Server facts

Local

Transport: stdio
Last checked: 2026-10-28

Stars: 1
Tools: —
Uptime: —

Tools

No tool list yet. Remote servers get one after the first health check.

FAQ

Questions

How do I install Dbeaver?

Copy the stdio config for Cursor, Claude Desktop, Claude Code, or VS Code on this page. It uses the publisher's install, not a PlugMCP proxy.

Does Dbeaver need an API key?

No required environment variable is listed on this record.

What transport does Dbeaver use?

Dbeaver runs locally over stdio.

When was Dbeaver last checked?

Last checked 2026-10-28. DBeaver MCP Server is an Eclipse/OSGi plugin that runs inside DBeaver and exposes configured database connections to AI agents through a local MCP Streamable HTTP server.

From the publisher

Documentation

DBeaver MCP Server

DBeaver MCP Server is an Eclipse/OSGi plugin that runs inside DBeaver and exposes configured database connections to AI agents through a local MCP Streamable HTTP server.

The plugin uses DBeaver's connection registry, drivers, and live sessions directly. It does not read or decrypt credentials-config.json.

Install From GitHub Pages

In DBeaver, open Help -> Install New Software....

Add the following update site:

https://bachden.github.io/dbeaver-mcp-server/

Select DBeaver MCP Server and complete the installation wizard.
Restart DBeaver when prompted.

This URL is a p2 repository for DBeaver. The root may not render an HTML page in a browser; DBeaver reads p2.index, artifacts.xml.xz, and content.xml.xz directly.

Human In The Loop

The plugin follows a what you see is what you get rule:

list_connections lists both open and closed connections and includes a connected field.
get_schema, run_query, and execute_sql only use connections that are already open in DBeaver.
The plugin never opens a database connection automatically or presents a credential prompt on the user's behalf.
When a connection is closed, the tool returns an error and asks the user to open it manually in DBeaver first.
Tunnels, credentials, MFA, and the decision to connect remain under the user's control in the DBeaver UI.

Tools

Tool	Description
`list_connections`	Lists connections, their `connectionId`, endpoint, and `connected` state.
`get_schema`	Browses catalogs, schemas, tables, and columns through a dotted `path`.
`run_query`	Runs read-only SQL such as SELECT, WITH, SHOW, EXPLAIN, and DESCRIBE.
`execute_sql`	Runs any SQL, including INSERT, UPDATE, DELETE, and DDL. It can modify or delete data.

All three database-access tools reject closed connections.

Server And UI

The default endpoint is:

POST http://127.0.0.1:8722/mcp

The DBeaver MCP menu provides:

Start Server
Stop Server
Connections...
Settings...

The Window -> Preferences -> DBeaver MCP Server page configures the listener host, port, bearer-token authentication, and client snippets for Claude Code, Claude Desktop, Codex CLI, or raw HTTP.

The server supports initialize, ping, tools/list, tools/call, and JSON-RPC batches. It does not provide an SSE stream.

Project Layout

dbeaver-mcp-server-plugin-parent
├── pom.xml
├── .github/workflows/
├── dbeaver-mcp-server-plugin
│   ├── META-INF/MANIFEST.MF
│   ├── plugin.xml
│   ├── dbeaver-ci.target
│   ├── pom.xml
│   └── src/
├── dbeaver-mcp-server-plugin.feature
│   ├── feature.xml
│   └── pom.xml
└── dbeaver-mcp-server-plugin.updatesite
    ├── category.xml
    └── pom.xml

The modules are:

dbeaver-mcp-server-plugin-parent: the Maven/Tycho reactor.
dbeaver-mcp-server-plugin: the OSGi bundle.
dbeaver-mcp-server-plugin.feature: the installable p2 feature.
dbeaver-mcp-server-plugin.updatesite: the p2 repository.

Development Setup

Requirements:

DBeaver 23+ installed.
Java 21.
Eclipse IDE with PDE and m2e.
Maven 3.9 or newer.

Import the repository root with File -> Import -> Existing Maven Projects. Eclipse will discover the four reactor projects.

Build The p2 Update Site

Run from the repository root:

mvn clean verify

Tycho builds the bundle, feature, and p2 repository. The complete repository is copied to:

dist/
├── artifacts.jar
├── content.jar
├── features/
├── plugins/
└── p2.index

Install From A Local Update Site

Build the project to create dist/.
In DBeaver, open Help -> Install New Software....
Select Add... -> Local... and choose the dist/ directory.
Give the repository a name such as Local DBeaver MCP.
Select DBeaver MCP Server and complete the wizard.
Restart DBeaver when prompted.

There is no need to copy JARs, edit bundles.info, or use dropins/. Normal installation and update flows do not require -clean.

Update The Plugin

For local development:

Rebuild the update site:
```
mvn clean verify
```
In DBeaver, select Help -> Check for Updates.
Install the new version and restart DBeaver when prompted.

If the Error Log reports Another singleton bundle selected after the exact same version has been reinstalled, close DBeaver completely and run it once with:

/Applications/DBeaver.app/Contents/MacOS/dbeaver -clean

DBeaver can be started normally after that one clean launch.

GitHub Actions And Pages

The repository contains two workflows:

Build and Publish DBeaver Update Site runs on pushes to main, builds with the shared dbeaver-ci.target, and deploys dist/ to GitHub Pages.
Validate DBeaver Update Site runs for pull requests and uploads dist/ as a build artifact.

The published update site is:

https://bachden.github.io/dbeaver-mcp-server/

Do not commit target/ or dist/; Maven recreates these artifacts in CI.

Connect An AI Agent

Claude Code example with authentication disabled:

claude mcp add --transport http dbeaver http://127.0.0.1:8722/mcp

When authentication is enabled, clients must send:

Authorization: Bearer <token>

Codex CLI example:

[mcp_servers.dbeaver]
url = "http://127.0.0.1:8722/mcp"

[mcp_servers.dbeaver.http_headers]
Authorization = "Bearer <token>"

Security

The server binds to 127.0.0.1 by default.
Enable bearer-token authentication when binding to a LAN address or 0.0.0.0.
execute_sql accepts unrestricted SQL and can modify or delete data.
The run_query filter is not a replacement for database permissions. Use a read-only database account when a hard write restriction is required.
The user must open a connection in DBeaver before an agent can access that database.
The server can be stopped at any time from the menu or status bar.