Skip to main content
Conduktor Console includes a Model Context Protocol (MCP) server. Connect an AI assistant, such as Claude Desktop, to it and ask questions about your Kafka platform in plain language: which topics a team owns, why a consumer group is lagging, who changed a topic’s retention last week or how much storage each application uses.
From our blog: Conduktor MCP and skills for AI agents How Conduktor’s native MCP server connects AI assistants to Kafka with built-in RBAC and no separate server to deploy.

​
Why use the Conduktor MCP server

  • Nothing extra to deploy: the MCP server is part of Console, so there’s no separate service to host, secure or upgrade.
  • Your permissions, not more: every request runs with the permissions of your own token, so the assistant only sees what you can see in Console.
  • One view of the whole platform: tools cover Kafka resources, Schema Registry, Kafka Connect, access control, Self-service, Chargeback, the audit log and Insights.
  • Read-only by default: all tools are currently read-only except set-topic-labels, which adds or updates labels on a topic.
  • Built for large clusters: tools filter, aggregate and page on the server, so the assistant can answer questions about thousands of topics without pulling every one into the conversation.

​
What you can do with it

The results of using MCP with various AI providers and models are open to interpretation by the AI, and outside of Conduktor’s direct control.

​
Get started

Connect an AI assistant

Generate a token and connect Claude Desktop or another MCP client to Console.

Tools reference

See every tool the MCP server offers, and how to analyze topics with aggregate-topics.