Axual MCP Server

Bring your own AI. Keep control of your data.

Connect any AI agent to the Axual MCP. Create topics, deploy KSML or Flink SQL jobs and browse messages in plain language, with exactly your permissions and nothing more.
Read docs
Watch tutorial
Create a Kafka topic orders with 30 days retention in dev environment.
Successfully created Kafka topic orders in dev environment.

The topic is now ready to use!
Why Axual MCP

AI agents are only as safe as the data they can reach

Use the AI tools you already have

‍

Any MCP-compatible client, including EU-hosted or self-hosted models. No new AI to adopt.

‍

Your permissions, nothing more

‍

The agent acts as you, through the Self-Service APIs. No elevation, no direct Kafka access, ownership and approvals still apply.

Runs where your platform runs

‍

On-prem, in your own cloud or on Axual Cloud. Your data doesn’t go anywhere new.

Who is it for

Built for everyone who works with Kafka

For developers

Build faster on Axual.
  • Discover topics and environments
  • Create topics with correct ownership
  • Generate realistic test data using KSML
  • Deploy streaming applications with KSML or Flink SQL
  • Debug by browsing real messages

For platform engineers

Control everything through Axual.
  • Enforce environments (dev, staging, production)
  • Let teams use AI with the same guardrails as everything else
  • Manage ownership and groups
  • Standardize Kafka workflows
  • Enable safe self-service without exposing raw Kafka

For engineering leaders

Adopt AI without losing oversight.
  • Reduce onboarding time
  • Increase developer productivity
  • Maintain governance across teams
  • Roll out AI across teams without new risk

How Axual MCP server works

Built for Governance
MCP Capabilities
Generating KSML apps with AI

Built for Governance

  • Uses Axual Self-Service APIs
  • Enforces environments and ownership groups
  • Supports access requests and approval workflows
  • All actions are validated and governed
  • You describe a task in natural language
  • The AI selects an MCP tool
  • Axual MCP executes the request through Self-service APIs
  • The platform enforces ownership, environments and access
  • Results are returned in the same conversation
none
> Create a Kafka topic called 'orders' in the dev environment
none
< Choose the owner group for this topic
DataEngineering
Testers
PlatformAdmins
none
> DataEngineering
none
< Kafka topic orders created in dev environment owned by
none
> Produce 50 test messages to 'orders' topic as per schema
none
< 50 test messages produced to 'transactions' topic.

MCP Capabilities

  • Topics: create, search, manage
  • Deployments: inspect across environments
  • Data: produce and browse messages
  • Applications: create, search, manage
  • KSML: generate KSML apps with AI and register/run them on Axual
  • Flink SQL: write and deploy jobs from a plain-language description
  • Access: request and enforce Kafka permissions

How to generate Kafka streams with AI

  • Define Kafka Streams pipelines in YAML
  • Filter, transform, and route data
  • Generate realistic test data
  • Deploy applications directly through Axual
Kafka MCP use cases

What you can do with the Axual MCP server

Discover Axual resources

  • What topics does my team own?
  • What environments are available?
  • Where is this topic deployed?

Build and test

  • Create a topic orders in dev, owned by DataEngineering
  • Produce 50 realistic test messages to orders
  • Deploy a KSML app that filters pending transactions
  • Write a Flink SQL job that counts orders per minute

Troubleshoot

  • Why is no consumer reading from this topic?
  • Show messages with errors in the last hour

Govern and optimize

  • What topics are unused or misconfigured?
  • Which topics require access changes?
  • What should I fix before production?

Outcomes

Faster Axual Workflows

AI adoption without new risk

Safer self-service within governance boundaries

Faster onboarding for Kafka users

MCP security

Does Axual MCP meet your security team’s standards?

Same permissions, same boundaries: AI operates with the exact permissions of the user. No elevation, no hidden access.
No direct Kafka access: All actions go through Axual Self-Service APIs: governance, ownership, and approvals remain enforced.
The agent only sees what the user is allowed to see. Use an EU-hosted or self-hosted model to keep data inside your own environment.

Works with your IAM: Fully integrated with your Identity and Access Management system.

See what the Axual Server can do for your infrastructure

Book a demo
Read the documentation

Frequently asked questions

What is the Axual MCP Kafka server?

The Axual MCP Server connects AI assistants to the Axual platform, letting you manage Kafka using natural language instead of raw commands or UI clicks. It works through Axual's Self Service APIs, so every action such as creating topics, deploying apps, or browsing messages stays governed by existing ownership, environment, and access rules.

How does the Axual MCP Kafka server work?

You describe what you want in plain language, for example "create a topic called orders in dev," and the AI selects the right MCP tool to carry out the request. The server executes it through Axual's Self Service APIs, which enforce ownership, environment, and access policies before returning the result in the same conversation.

Is the Axual MCP Kafka server secure? Does it create new access for AI?

No, the MCP server doesn't grant any new access; it only makes a user's existing Axual permissions accessible via AI. The AI operates with the exact same permissions as the user (no elevation), never touches Kafka directly, and all actions pass through Axual's governed Self Service APIs, fully compatible with enterprise Identity and Access Management systems.

Can my AI agent create and deploy Flink SQL jobs?

Yes. Through the Axual MCP server, your AI assistant can create and manage Flink SQL applications. Describe the job in plain language, and the agent writes the SQL and deploys it as a managed job on a Flink cluster your tenant has registered. Flink SQL support is currently in preview.

Do I need to write Flink table definitions myself?

No. Axual can generate the table definitions from your connected Kafka topics, so you or your agent only need to write the transformation logic. You can still write the tables by hand if you prefer.

What can I do with the Axual MCP Kafka server?

You can discover and search topics, environments, and deployments; create topics with correct ownership; produce and browse test or real messages; create and manage streaming applications; generate and deploy KSML apps; and request or enforce Kafka access permissions, all through natural language prompts.

What is KSML and how does the Axual MCP server use it?

KSML (Kafka Streams Markup Language) lets you define Kafka Streams pipelines in YAML to filter, transform, and route data without writing code. Through the MCP Server, you can ask an AI assistant to generate a KSML application, register it, and deploy it directly on Axual, for example filtering pending transactions from a topic into another one.

Should I use KSML or Flink SQL?

Both run as managed applications on Axual. KSML suits teams that want to define Kafka Streams pipelines in YAML with Python logic, without running extra infrastructure. Flink SQL suits teams that already know SQL or run Flink, and want to process streams on their own Flink cluster. Your agent can work with either.

How do I connect a client to the Axual MCP server?

First, get your MCP server URL from Axual Self Service. Then connect your AI client. For example, with Claude Code you'd run claude mcp add --transport http axual [your-mcp-server-url]. Once connected, you can start asking questions like "What environments are available in Axual?"