MCP Kafka Server

A Kafka tool based on the MCP protocol, supporting querying, reading/writing, and replaying Kafka messages

Developer tools Monitoring #Kafka Tool #Message Replay #MCP Protocol .Python

rating : 2 points

downloads : 13

update time : 2025-04-28

What is the MCP Kafka Tool?

The MCP Kafka Tool is an intelligent Kafka management interface that enables management operations on Kafka clusters through natural language interaction. It encapsulates complex Kafka command-line operations, allowing users to complete professional data pipeline management tasks with simple instructions.

How to use the MCP Kafka Tool?

You can use it in two ways: 1) Run the script directly in the local STDIO mode. 2) Provide an HTTP interface through the web service mode. It supports seamless integration with MCP clients such as Claude and Cherry Studio.

Applicable Scenarios

Scenarios such as real-time data monitoring, historical data analysis, Kafka operation and maintenance management, data pipeline debugging, and event backtracking. It is especially suitable for users who need to interact with Kafka frequently but do not want to memorize complex commands.

Main Features

Topic ManagementView all available topics and their detailed information, including partition status and message volume statistics

Data OperationsSupport sending messages to topics (JSON format is supported) and reading the latest messages

Time ReplayReplay historical messages within a specified time range, supporting a time window accurate to the second

Advantages and Limitations

Advantages

No need to memorize complex Kafka commands, interact through natural language

Provide a visual time selector for historical data replay

Support both local and remote access modes

Ready to use with mainstream MCP clients

Limitations

Requires Python environment support

Currently only supports single-node Kafka connection configuration

Limited performance when processing large message volumes

How to Use

Install Dependencies

Ensure that Python 3.7+ is installed, and then run pip to install the required dependencies

Select the Running Mode

Select the local STDIO mode or the web service mode according to your needs

Connect to the Client

Connect to the local or remote service endpoint in an MCP client (such as Claude)

Usage Examples

Monitor Real-time Data StreamsDevelopers need to monitor new messages in the order topic in real-time

Troubleshoot FaultsOperation and maintenance personnel need to check error logs within a specific time period

Frequently Asked Questions

How to modify the Kafka connection address?

Which Kafka versions are supported?

What is the maximum number of messages that can be processed?

Related Resources

Kafka Official Documentation

Official technical documentation for Apache Kafka

MCP Protocol Specification

Official specification of the Model Context Protocol

Example Code Repository

Contains more usage examples and extended functions

🚀 MCP Kafka Tool

This is a Kafka tool based on the MCP (Model Context Protocol). It enables the connection to Kafka through large models and offers the following capabilities:

Query information about specified topics.
Read from and write to specified topics.
Replay data from specified topics within a given time frame.

📦 Installation

pip install -r requirements.txt

💻 Usage Examples

Local Mode (Standard Input/Output)

Run the mcp_kafka_server.py file directly:

python mcp_kafka_server.py

Then connect to the local MCP server in an MCP client (e.g., Claude, Cherry Studio).

Web Service Mode (SSE)

Run the mcp_kafka_sse_server.py file to start the web service:

python mcp_kafka_sse_server.py

The service will start at http://localhost:8000, and the MCP endpoint is http://localhost:8000/mcp. Then connect to this remote MCP server in an MCP client.

✨ Features

1. List All Topics

list_topics()

Returns a list of all available topics in Kafka.

2. Get Topic Information

get_topic_info(topic_name)

Retrieves detailed information about the specified topic, including the number of partitions, the start and end offsets of each partition, and the number of messages.

3. Send a Message

send_message(topic_name, message, key)

Sends a message to the specified topic. The message can be a JSON string or a regular string, and the key is an optional parameter.

4. Read Messages

read_messages(topic_name, max_messages, timeout_ms)

Reads the latest messages from the specified topic. The max_messages parameter specifies the maximum number of messages to read (default is 10), and the timeout_ms parameter specifies the timeout in milliseconds (default is 1000).

5. Replay Historical Messages

replay_messages(topic_name, start_time, end_time, max_messages)

Replays messages within the specified time frame. The start_time and end_time are ISO-formatted time strings (e.g., '2023-01-01T00:00:00'), and the max_messages parameter specifies the maximum number of messages to replay (default is 100).

🔧 Technical Details

The tool is configured to connect to localhost:9092 by default. To modify the connection address, change the bootstrap_servers parameter in the initialization of the KafkaManager class.