Open Source · Apache License 2.0

Stop managing Kafka with tickets and scripts.

Resource as Code for Apache Kafka Apache Iceberg and beyond.

Declare topics, ACLs, schemas, quotas, and connectors as YAML. Review in pull requests. Jikkou reconciles your Git repository against your real clusters. No state file, no drift, no tickets.

Get Started → View on GitHub

5+Kafka platforms

20+resource types

CLI + APIdual interface

Apache-2.0license

kafka-topic.yaml YAML

apiVersion: kafka.jikkou.io/v1
kind: KafkaTopic
metadata:
  name: 'orders-events'
spec:
  partitions: 6
  replicas: 3
  configs:
    retention.ms: 604800000
    min.insync.replicas: 2

Terminal

$ jikkou diff --files kafka-topic.yaml
+ CREATE  KafkaTopic/orders-events (partitions: 6, replicas: 3)

$ jikkou apply --files kafka-topic.yaml
✓ Validating resources…
✓ Computing changes…
CHANGED  KafkaTopic/orders-events

✓ Reconciliation completed. 1 changed, 0 failed

Who it's for

Built for the team that runs Kafka for everyone else

Platform teams

50 app teams, and every topic request is a ticket.

Give teams a paved path: YAML + pull request + CI. You keep the guardrails: naming rules, replication minimums, and partition limits, enforced on every change.

DevOps & SRE

Nobody remembers why that topic has 96 partitions.

Your whole Kafka configuration, versioned in Git and diffed against reality. Out-of-band changes show up; history explains itself.

App teams

We just want to ship, not learn Kafka administration.

Self-service topics and schemas inside platform-enforced validations. A pull request instead of a ticket queue.

Workflow

Describe. Diff. Apply.

A reconciliation loop, not a script: Jikkou computes the difference between your Git repository and the real cluster, and applies only what changed.

1 · YAML

Describe

Define topics, ACLs, schemas, quotas, and connectors in declarative YAML, with Jinja templating for the 300 topics that differ by one value.

2 · jikkou diff

Diff

Preview every change against the actual cluster state, in the pull request, before anything happens. Validations catch policy violations here.

3 · jikkou apply

Apply

Reconcile the desired state. Only the necessary changes are applied: created, altered, deleted, reported.

Why Jikkou

The parts that actually change your week

⇄

Stateless reconciliation & drift detection

No state file to protect or migrate. Every diff compares Git against the real cluster, so the topic someone "fixed" by hand during an incident shows up tomorrow morning instead of at the next audit.

⛁

Multi-cluster fleet apply

Provider groups target your whole estate (on-prem, Confluent Cloud, Aiven, MSK) with one command. Same YAML, five clusters, zero copy-paste.

{}

Templating at scale

Jinja templates + per-environment values files. One definition for the 300 topics that differ by a name and a number.

⎇

GitOps-native

YAML in Git, diff in the pull request, apply in CI. Review, audit trail, and rollback come free with your existing workflow.

◎

Dry-run mode

See exactly what would happen. Nothing touches production unreviewed.

✓

Validation policies

Naming, min ISR, and partition limits, enforced before anything is applied.

▸_

CLI + REST API server

Native binaries for CI; an API server for team-wide automation.

＋

Extensible providers

A Java API to manage any resource, not just Kafka.

Coverage

One model, every platform

Apache Kafka Confluent Cloud Amazon MSK Aiven Redpanda Schema Registry Kafka Connect AWS Glue

Beyond Kafka

The engine is platform-agnostic

Apache Iceberg tables, views, and namespaces: same model, same workflow, same pull request. Your platform's next system can be a provider away.

apiVersion: iceberg.jikkou.io/v1
kind: IcebergTable
metadata:
  name: 'analytics.orders'
spec:
  schema:
    fields:
      - name: 'order_id'
        type: long

Honest comparisons

Already using Terraform or Strimzi?

Good news: they're not enemies. Here is exactly where each tool wins, and where Jikkou fits.

Jikkou vs Terraform

Terraform provisions your cluster. Jikkou manages what lives inside it.

No state file to drift: every diff runs against the real cluster. Keep Terraform for infrastructure (clusters, networks, credentials).

Read the comparison →

Jikkou vs Strimzi Topic Operator

Keep the Topic Operator for topics. Jikkou covers everything it can't reach.

Schema Registry, Connect, ACLs, quotas, Confluent Cloud, Aiven, MSK, and every cluster outside your Kubernetes estate.

Read the comparison →

Community

Built in the open

20+contributors

8resource providers

v1.xstable and actively maintained

Running Jikkou in production? Tell us: your story helps the project more than anything. Real-world usage reports guide the roadmap and help other teams adopt with confidence.

Share on GitHub Discussions →

Your next topic is a pull request away.

Install in one command. Export your existing cluster in ten minutes.

$ sdk install jikkou

Read the docs → Join Slack GitHub