# Conduktor [Health: Active]

**Category:** 💻 Developer Tools  
**Repository:** https://github.com/conduktor/kafka-mcp  
**GitHub Stars:** 5  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/conduktor

## Description
Read-only Kafka access for AI assistants: topics, consumer groups, schemas, cluster health.

## Tools
Capabilities this server exposes over MCP:

- **list-clusters** — Every Kafka cluster this Console manages, with its Schema Registry, Connect clusters and flavor
- **get-cluster** — One cluster's registration by slug
- **insights-cluster** — Health score, topic and partition counts, serialization breakdown — the shape of a cluster in one call
- **list-topics** — Topic catalogue: names, labels, descriptions, partitions, replication, configs
- **get-topic** — One topic by exact name
- **list-topics-with-usage** — Topics with message count, size and throughput
- **query-topics** — Filter topics across a cluster
- **aggregate-topics** — Group topics and compute a statistic per group — count, sum, average, min, max
- **insights-topics** — Partition skew, replication problems, and what is quietly wrong
- **get-last-messages** — The last N messages from a topic
- **get-record-at** — A specific record, by partition and offset — the one blocking a consumer
- **list-consumer-groups** — Groups with state, lag and member count
- **get-consumer-group** — One group in detail
- **list-consumer-groups-by-topic** — Who actually reads this topic
- **get-schema-version** — A specific version
- **check-schema-compatibility** — Whether a change breaks consumers, before it ships
- **list-interceptors** — Gateway interceptors configured on a cluster
- **list-acl-bindings** — Who is allowed to do what
- **list-certificates** — Certificates and their expiry
- **list-applications** — Applications registered in the catalogue
- **get-stream-lineage** — What flows into what
- **query-audit-log** — Who did what, when
- **get-chargeback-report** — Cost attributed per team

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `npx` (confidence: high):

```json
"mcpServers": {
  "conduktor": {
    "command": "npx",
    "args": ["-y","@conduktor/mcp"],
    "env": {
      "CONDUKTOR_CONSOLE_URL": "",
      "CONDUKTOR_API_TOKEN": ""
    }
  }
}
```

**Requires environment variables:** `CONDUKTOR_CONSOLE_URL`, `CONDUKTOR_API_TOKEN` — the values above are empty placeholders; fill in real credentials before running (see the repository for what each one is for).

## Documentation

## What Conduktor MCP server does

The Conduktor MCP server gives AI assistants an MCP interface to Kafka estates registered in Conduktor Console. It covers both Kafka resource inspection and related Console control-plane metadata, including cluster registrations, topic catalogues, consumer groups, schemas, certificates, applications, permissions, lineage, audit activity, and team chargeback information.

The supplied tools are primarily read-oriented. They can show cluster health, topic usage, partition and replication issues, consumer lag, recent records, records at specific offsets, schema versions, and compatibility results. This makes the server suitable for investigation and operational questions where the assistant needs current information from Console rather than static documentation.

## How it works

The server endpoint runs inside Conduktor Console. A client launches the published `@conduktor/mcp` package, which connects to the Console MCP endpoint; the package appends `/api/mcp` to the configured Console URL when needed. The repository also documents a direct `mcp-remote` setup for clients that do not use the package wrapper.

Each request carries a Console Personal Access Token. Console RBAC determines which clusters and resources the caller can access, and calls are recorded in the Console audit log. The material describes the model-facing data as control-plane metadata, while Kafka records remain in Kafka unless a task uses a tool that reads them. Message inspection tools such as `get-last-messages` and `get-record-at` are exceptions that can retrieve record data.

## Setup and configuration

Create a Personal Access Token in Conduktor Console, then configure an MCP client with the launcher:

```json
{
  "mcpServers": {
    "conduktor": {
      "command": "npx",
      "args": ["-y", "@conduktor/mcp"],
      "env": {
        "CONDUKTOR_CONSOLE_URL": "https://console.example.com",
        "CONDUKTOR_API_TOKEN": "your-personal-access-token"
      }
    }
  }
}
```

`CONDUKTOR_CONSOLE_URL` identifies the Console instance, and `CONDUKTOR_API_TOKEN` supplies authentication. URLs that already end with `/api/mcp` are accepted. As an alternative, invoke `mcp-remote` directly and pass the token as a Bearer authorization header.

The repository includes a demo environment that starts Kafka, Console, and the MCP endpoint with Docker Compose. It is seeded with topics, traffic, and a lagging consumer group for evaluation without an existing Kafka estate.

## Tools and capabilities

The Conduktor MCP server exposes tools in these areas:

- **Clusters:** list managed clusters, retrieve one by slug, and inspect health, topic counts, partition counts, and serialization breakdown.
- **Topics:** browse, retrieve, filter, aggregate, and inspect usage, including message counts, size, throughput, partition skew, and replication problems.
- **Messages:** retrieve the latest messages or locate a record by partition and offset.
- **Consumer groups:** inspect state, lag, members, and the groups consuming a particular topic.
- **Schemas:** retrieve a schema version and test compatibility before a change is introduced.
- **Governance and operations:** inspect interceptors, ACL bindings, certificate expiry, registered applications, stream lineage, audit events, and team chargeback reports.

## Limitations and notes

The server depends on Conduktor Console and an authorized Personal Access Token; it is not a standalone Kafka deployment or a replacement for Console. Access is limited by the token holder's Console RBAC permissions, so an assistant cannot see resources unavailable to that identity.

The supplied listing describes the current use case as read-only Kafka access. Although the repository documentation discusses a broader agent model and additional operational workflows, the tools listed for this entry should be treated as the supported interface when evaluating coverage.

The endpoint runs inside Console, so there is no separate server deployment described for production use. For launcher or repository issues, the README distinguishes those from issues with the server running in Console, which should go through Conduktor support.

_Full upstream README: https://allmcps.com/mcp/conduktor/readme_