MCP Confluent

What confluentinc/mcp-confluent MCP server does

The confluentinc/mcp-confluent MCP server connects MCP-compatible assistants to Confluent Cloud, Confluent Platform, and standalone Apache Kafka deployments. Its toolset spans operational Kafka tasks, Flink SQL, schema management, Kafka Connect, catalog metadata, and product documentation. It also includes tools that explain disabled capabilities and show which connections and tools are active.

The project is open source and intended for use with clients such as Claude Desktop, Claude Code, Cursor, VS Code, Goose, and Gemini CLI. The README specifically identifies Claude Desktop and Cursor among the supported client options.

How it works

Configuration is resolved from a YAML file. Service blocks in that configuration determine which groups of tools are enabled, so a deployment does not need to expose every capability. A bare configuration can still provide documentation search and configuration-diagnostic tools.

For Confluent Cloud, the server communicates with configured service endpoints and authenticated APIs. The Kafka, Flink, connector, Schema Registry, and catalog groups marked in the documentation can also use Confluent Cloud OAuth, allowing browser-based sign-in instead of API keys. The same project can target Confluent Platform and self-managed Kafka, which is a stated reason to choose it over Confluent's separately managed Cloud MCP offering.

Setup and configuration

The confluentinc/mcp-confluent MCP server requires Node.js 22.19.0 or newer. Generate a starter configuration with npx @confluentinc/mcp-confluent --init-config, then edit the resulting config.yaml with the connection details for the systems you want to expose. Start the server with npx @confluentinc/mcp-confluent --config ./config.yaml.

The repository's example configuration and configuration guide are the sources for the available service blocks and connection variables. Use --list-tools to inspect the tools made available by the resolved configuration. A Confluent Cloud account is needed when connecting to Confluent Cloud.

Tools and capabilities

The confluentinc/mcp-confluent MCP server includes tools for:

• Listing, creating, deleting, producing to, and consuming from Kafka topics.
• Inspecting consumer groups, lag, and topic configuration.
• Creating, listing, inspecting, and deleting Flink SQL statements, plus retrieving results and exceptions.
• Browsing Flink catalogs, databases, tables, and table schemas.
• Checking Flink statement health, detecting issues, and viewing profiles.
• Listing, inspecting, creating, deleting, pausing, resuming, restarting, and updating Kafka Connect connectors.
• Reviewing connector offsets, status, tasks, logs, errors, and recommendations.
• Listing, creating, and deleting schemas.
• Searching topics by name or tag and managing entity tags.
• Searching Confluent product documentation and retrieving complete documentation pages.
• Listing configured connections, explaining unavailable tools, and generating configuration guidance.

Limitations and notes

The open-source project is community-supported. The README states that Confluent does not provide dedicated support, service-level commitments, or guaranteed response times for this server. Confluent Cloud customers may instead use a separately managed MCP server, but this open-source option is the one documented here for Confluent Platform, self-managed Kafka, and customization or extension of the toolset.