About mcp-kafka
mcp-kafka is an MCP server for monitoring and managing Apache Kafka clusters from MCP clients. It starts in read-only mode and governs operations through access levels, topic scope, and deletion controls.
Key features
- 01Inspect clusters, topics, and consumer groups
- 02Check per-partition and total consumer lag
- 03Expose management tools by access level
- 04Restrict topics with allowlists and protections
- 05Gate deletes, dry-run, and audit operations
Use cases
- 01Check which consumer groups have lag
- 02Inspect topic metadata and offsets
- 03Create topics or adjust partitions
- 04Manage Kafka clusters within defined limits
MCP support
The server exposes Kafka monitoring and management tools to MCP clients. Tools are registered according to read-only, read-write, or admin mode, and MCP elicitation can request approval for high-impact actions.
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
npx -y @dockndevai/mcp-kafkaKAFKA_BROKERSComma-separated broker list (host:port).KAFKA_MODEAccess mode: read-only | read-write | admin.Local servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.
Tools
6Saved tool list · updated 3d ago
This tool takes no parameters.
Details
From the project READMEmcp-kafka
A Model Context Protocol server for Apache Kafka. It lets an MCP-capable client (Claude Desktop, Claude Code, etc.) monitor and manage Kafka clusters — topics, partitions, configs, and consumer groups (including lag) — with behaviour controlled entirely by flags.
Safe by default: it starts read-only, can be scoped to an allowlist of topics, protects internal/critical topics from mutation, and gates destructive operations behind an explicit opt-in.
Features
- Monitoring — cluster/broker info, topic metadata and offsets, consumer groups, and per-partition + total consumer lag.
- Management — create topics, add partitions, alter topic configs, reset group offsets; delete topics/groups (admin).
- Access modes —
read-only→read-write→admin, layered so a mode never exposes tools above its level. - Security flags — topic allowlist, protected/internal topics, delete gating, dry-run, and JSON audit logging (see below).
- Auth — plaintext, TLS, and SASL (PLAIN / SCRAM-SHA-256 / SCRAM-SHA-512).
Security model
| Concern | Flag | Default | Effect |
|---|---|---|---|
| What can the server do? | KAFKA_MODE | read-only | read-only exposes only monitoring; read-write adds management; admin adds deletes. Tools above the mode are never registered. |
| Which topics are in scope? | KAFKA_TOPIC_ALLOWLIST | (all) | When set, operations on other topics are refused. |
| Protect internal topics | KAFKA_PROTECT_INTERNAL_TOPICS | true | Topics starting with _ can be read but never mutated. |
| Protect specific topics | KAFKA_PROTECTED_TOPICS | (none) | Additional read-only-forever topics. |
| Can it delete? | KAFKA_ALLOW_DELETE | false | delete_topic / delete_consumer_group need this and admin mode. |
| Preview without touching the cluster | KAFKA_DRY_RUN | false | Write/admin tools validate + log intent, then return. |
| Audit trail | KAFKA_AUDIT_LOG | true | Emits a JSON line to stderr per guarded operation. |
| Interactive confirmation | (automatic) | — | Destructive & high-impact actions prompt the human to approve via MCP elicitation before running; clients without elicitation fall back to the *_ALLOW_* gate. |
Tools
Read (read-only+): cluster_info, list_topics, describe_topic, topic_offsets, list_consumer_groups, describe_consumer_group (with lag)
Write (read-write+): create_topic, create_partitions, alter_topic_config, reset_consumer_group_offsets
Admin (admin): delete_topic, delete_consumer_group (both need KAFKA_ALLOW_DELETE)
Quickstart — add to your agent
Published on npm as @dockndevai/mcp-kafka. No clone or build needed — your MCP client runs it on demand with npx. Start in read-only mode; see .env.example for every variable and docs/CLIENTS.md for the full per-client guide.
Claude Code (CLI)
claude mcp add kafka -e KAFKA_BROKERS="localhost:9092" -e KAFKA_MODE="read-only" -- npx -y @dockndevai/mcp-kafka
Claude Desktop · Cursor · Windsurf — same block in claude_desktop_config.json, .cursor/mcp.json, or ~/.codeium/windsurf/mcp_config.json:
{
"mcpServers": {
"kafka": {
"command": "npx",
"args": [
"-y",
"@dockndevai/mcp-kafka"
],
"env": {
"KAFKA_BROKERS": "localhost:9092",
"KAFKA_MODE": "read-only"
}
}
}
}
OpenAI Codex CLI — in ~/.codex/config.toml:
[mcp_servers.kafka]
command = "npx"
args = ["-y", "@dockndevai/mcp-kafka"]
env = { KAFKA_BROKERS = "localhost:9092", KAFKA_MODE = "read-only" }
VS Code (GitHub Copilot, Agent mode) — in .vscode/mcp.json:
{
"servers": {
"kafka": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@dockndevai/mcp-kafka"
],
"env": {
"KAFKA_BROKERS": "localhost:9092",
"KAFKA_MODE": "read-only"
}
}
}
}
Example prompts
- "Which consumer groups have the most lag right now?"
- "Describe the
orderstopic and show its offsets." - "Create a topic
eventswith 6 partitions and 7-day retention." (needsread-write)
Run from source (development)
Prefer the published package above. To run from a clone:
npm install
npm run build
node dist/index.js # with the environment variables set
Develop
npm run dev
npm test
npm run typecheck
Publishing
This server ships a server.json for the official MCP registry and an mcpName for npm ownership validation. See PUBLISHING.md for publishing to npm and listing on the MCP registry, Smithery, Glama, Cursor, and PulseMCP.
License
MIT
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.