Kafka Dataops Mcp

io.github.AguantarPublished by GitHub account Aguantarv0.1.0Updated Oct 24, 2026

DataOps Kafka MCP server with consumer lag diagnosis and broker monitoring

VerifiedSTDIODesktop onlyDatabasesSecurity & Monitoring

Overview

AI-generated overview

A local Kafka DataOps MCP server that lets an assistant inspect consumer group lag, topics and broker health with incident-pattern diagnosis.

What you can do
Inspect consumer groups: read lag and get incident-pattern diagnosis hints for stalled pipelines, including Flink-related failure patterns and checkpoint warnings. Examine topics: view topic details with ISR and replication health checks, and list the topic catalog with built-in descriptions. Check the cluster: report brokers, the controller and under-replicated partitions.
When to use it
Useful when you operate Kafka or CDC pipelines and want an assistant to help triage consumer lag, replication problems or broker health during incidents and routine checks.
Requirements
Runs locally as a Python package installed with pip, launched over stdio. Needs network access to a Kafka cluster, configured through the KAFKA_BOOTSTRAP_SERVERS environment variable (default localhost:9092); KAFKA_COMMAND_TIMEOUT sets the command timeout in seconds.
Before you install
It connects to your Kafka cluster and reads operational metadata such as consumer groups, topics and broker state; point it only at clusters you are authorized to inspect. No authentication is declared, so access control depends on your cluster and network setup.

Installation

In SourceWeft

  1. Open Kafka Dataops Mcp in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Desktop only via STDIO. STDIO servers start a local process, so they need the SourceWeft desktop host.

Other MCP clients

Follow the launch instructions in the repository.

README

kafka-dataops-mcp

mcp-name: io.github.Aguantar/kafka-dataops-mcp

A DataOps-focused Kafka MCP server with consumer lag diagnosis and broker health monitoring. Diagnosis logic is based on actual CDC pipeline operational experience.

Features

  • kafka_consumer_lag — Consumer group lag with incident-pattern diagnosis
  • kafka_topic_info — Topic details with ISR/replication health checks
  • kafka_broker_status — Cluster health: brokers, controller, under-replicated partitions
  • kafka_list_topics — Topic catalog with built-in descriptions

Diagnosis based on real incidents

The diagnosis logic is not generic — it's based on actual operational experience:

  • Flink crash detection: "no active members" + growing lag = likely Flink Job failure (based on a 46-hour outage caused by MySQL DELETE → Debezium tombstone → Flink NPE)
  • Checkpoint vs consumer group: warns that Kafka consumer group reset alone is insufficient for Flink — checkpoints must be deleted first
  • ClusterIdMismatch: detects missing brokers and suggests Docker volume conflict as root cause
  • ISR monitoring: ISR < min.insync.replicas = write failures (critical)

Installation

bash
pip install kafka-dataops-mcp

Usage with Claude Code

Add to your .mcp.json:

json
{  "mcpServers": {    "kafka": {      "command": "kafka-dataops-mcp",      "env": {        "KAFKA_BOOTSTRAP_SERVERS": "localhost:9092"      }    }  }}

Environment Variables

VariableDefaultDescription
KAFKA_BOOTSTRAP_SERVERSlocalhost:9092Kafka bootstrap servers
KAFKA_COMMAND_TIMEOUT10Command timeout in seconds

License

MIT

Source: README.md at commit 6b30d7a

Tools

0
SourceWeft has not started this package in its sandbox yet, so its tools are not known.

Version history

1
  1. v0.1.0LatestSep 16, 2026

Score

50% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 days25 pts

    Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.

  • Auth declared10 pts

    Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.

  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts