consumerLABSCO SUMMARY
The whole Kafka CLI surface, without the CLI.
Most of what you do with Kafka day to day is create a topic, describe it, push a test message and read it back — this covers all four in one sentence each. Offset reset is in here too, which is the operation most worth being deliberate about: it changes where a live consumer group resumes.
What it is
A natural-language interface to a Kafka cluster. It covers both sides of the data path and the admin surface around it, so "publish this to test-kafka" and "list all topics" both become real Kafka operations.
What you get
producerpublishes a message to a topic;consumerreads messages back off onetopictools coverlist,create,deleteanddescribebrokerreturns broker informationpartitiontools return partitions and partition offsetsgroup_offsettools get and reset consumer group offsets
Requirements
Python with a virtual environment and the dependencies from requirements.txt. Two settings, in .env or exported: BOOTSTRAP_SERVERS pointing at your cluster, and MCP_TRANSPORT=stdio. Run python3 src/main.py, or point the client at that path directly. No account or key of its own — whatever your Kafka cluster requires is what it needs. Version 0.1.0.
Setup effort
Build from source — git clone https://github.com/shivamxtech/kafka-mcp.git
USE CASES
01
Testing a topic end to end
Publish a message and consume it back in the same conversation to confirm the pipe works.
02
Finding out what a topic actually looks like
Describe it and read its partitions and offsets instead of assembling CLI flags.
03
Getting a stuck consumer group moving
Read the group's current offsets, then reset them to where you want it to resume.
6 tools
Listed in project docs. Your client may show a different list.
producertopicbrokerpartitiongroup_offsetBEFORE YOU INSTALL
What to know first
Tools you get
6 tools named in the project's documentation. These six are groups (consumer, producer, topic, broker, partition, group offset); individual tool names are not documented.
Where it runs
The documented client runs the server on your machine.
INSTALLATION
Connect your client
Server setup
TERMINAL
git clone https://github.com/shivamxtech/kafka-mcp.git
cd kafka-mcp
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txtMCP Clients
Claude Desktop
CLAUDE DESKTOP · JSON
{
"mcpServers": {
"kafka-mcp": {
"command": "python3",
"args": [
"<YOUR_PATH_TO_KAFKA_MCP>/src/main.py"
],
"env": {
"BOOTSTRAP_SERVERS": "localhost:9092",
"MCP_TRANSPORT": "stdio"
}
}
}
}<YOUR_PATH_TO_KAFKA_MCP> is the folder you cloned. python3 has to be the interpreter you installed requirements.txt into — for the virtual environment above that is <YOUR_PATH_TO_KAFKA_MCP>/.venv/bin/python3.
