Confluent MCP Server

Name: Confluent MCP Server
Author: Confluent Inc.

Confluent's MCP server for Kafka topics, Flink SQL, connectors and Schema Registry, on Cloud or self-managed

Official · Confluent Inc. TypeScript Local · stdio Databases Cloud platforms DevOps and monitoring

GitHub npm Website

63 popularity score ★ 169 ↓ 1.9k/wk

Connect your client

Local (npm)

Add in one click

Add to Cursor Add to VS Code Add to LM Studio Add to Goose

Opens the app, which asks before it adds the server. Then replace each <NAME> in its settings with your value.

Or add it by command or config

FirstTerminal

npx @confluentinc/mcp-confluent --init-config

Claude CodeTerminal

claude mcp add confluent -- npx -y @confluentinc/mcp-confluent --config '<ABSOLUTE_PATH_TO_CONFIG_YAML>'

Claude Desktopclaude_desktop_config.json

{
  "mcpServers": {
    "confluent": {
      "command": "npx",
      "args": [
        "-y",
        "@confluentinc/mcp-confluent",
        "--config",
        "<ABSOLUTE_PATH_TO_CONFIG_YAML>"
      ]
    }
  }
}

Cursor~/.cursor/mcp.json

{
  "mcpServers": {
    "confluent": {
      "command": "npx",
      "args": [
        "-y",
        "@confluentinc/mcp-confluent",
        "--config",
        "<ABSOLUTE_PATH_TO_CONFIG_YAML>"
      ]
    }
  }
}

VS Code.vscode/mcp.json

{
  "servers": {
    "confluent": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "@confluentinc/mcp-confluent",
        "--config",
        "<ABSOLUTE_PATH_TO_CONFIG_YAML>"
      ]
    }
  }
}

Codex~/.codex/config.toml

[mcp_servers.confluent]
command = "npx"
args = ["-y", "@confluentinc/mcp-confluent", "--config", "<ABSOLUTE_PATH_TO_CONFIG_YAML>"]

Gemini CLITerminal

gemini mcp add --scope user confluent npx -- -y @confluentinc/mcp-confluent --config '<ABSOLUTE_PATH_TO_CONFIG_YAML>'

Local (from source)

Add in one click

Add to Cursor Add to VS Code Add to LM Studio

Opens the app, which asks before it adds the server. Then replace each <NAME> in its settings with your value.

Or add it by command or config

FirstTerminal

git clone https://github.com/confluentinc/mcp-confluent.git
cd mcp-confluent
pnpm install
pnpm run build

Claude CodeTerminal

claude mcp add confluent -- node '<ABSOLUTE_PATH_TO_CLONE>/dist/index.js' --config '<ABSOLUTE_PATH_TO_CONFIG_YAML>'

Claude Desktopclaude_desktop_config.json

{
  "mcpServers": {
    "confluent": {
      "command": "node",
      "args": [
        "<ABSOLUTE_PATH_TO_CLONE>/dist/index.js",
        "--config",
        "<ABSOLUTE_PATH_TO_CONFIG_YAML>"
      ]
    }
  }
}

Cursor~/.cursor/mcp.json

{
  "mcpServers": {
    "confluent": {
      "command": "node",
      "args": [
        "<ABSOLUTE_PATH_TO_CLONE>/dist/index.js",
        "--config",
        "<ABSOLUTE_PATH_TO_CONFIG_YAML>"
      ]
    }
  }
}

VS Code.vscode/mcp.json

{
  "servers": {
    "confluent": {
      "type": "stdio",
      "command": "node",
      "args": [
        "<ABSOLUTE_PATH_TO_CLONE>/dist/index.js",
        "--config",
        "<ABSOLUTE_PATH_TO_CONFIG_YAML>"
      ]
    }
  }
}

Codex~/.codex/config.toml

[mcp_servers.confluent]
command = "node"
args = ["<ABSOLUTE_PATH_TO_CLONE>/dist/index.js", "--config", "<ABSOLUTE_PATH_TO_CONFIG_YAML>"]

Gemini CLITerminal

gemini mcp add --scope user confluent node -- '<ABSOLUTE_PATH_TO_CLONE>/dist/index.js' --config '<ABSOLUTE_PATH_TO_CONFIG_YAML>'

Settings 9

`KAFKA_API_KEY`	Optional: Kafka API key, read into config.yaml through ${KAFKA_API_KEY}; with KAFKA_API_SECRET.
`KAFKA_API_SECRET`	Optional: Secret for the Kafka API key.
`SCHEMA_REGISTRY_API_KEY`	Optional: Schema Registry API key for config.yaml interpolation.
`SCHEMA_REGISTRY_API_SECRET`	Optional: Secret for the Schema Registry API key.
`CONFLUENT_CLOUD_API_KEY`	Optional: Confluent Cloud API key for the confluent_cloud block.
`CONFLUENT_CLOUD_API_SECRET`	Optional: Secret for the Confluent Cloud API key.
`DO_NOT_TRACK`	Optional: Set to true to stop usage data and Sentry error reports.
`NODE_EXTRA_CA_CERTS`	Optional: CA bundle path for brokers or Schema Registry behind an internal CA.
`MCP_API_KEY`	Optional: API key clients must send when the http or sse transport is on.

What it does

MCP Confluent is a local Node.js server from Confluent that gives an assistant tools for Apache Kafka and the services around it. It works against Confluent Cloud, Confluent Platform and plain self-managed Kafka: it can list, create and delete topics, produce and consume records, inspect consumer groups and their lag, register and delete schemas, and, on Confluent Cloud, run Flink SQL statements, manage Kafka Connect connectors, tag topics, manage Tableflow, query operational metrics and read billing costs.

What it exposes depends on a config.yaml you pass with --config. Each connection in the file is either direct (bootstrap servers, Schema Registry endpoint and API keys per service) or oauth (a Confluent Cloud browser sign-in), and each service block you fill in switches on its group of tools. A file may hold several direct connections but at most one oauth one, and each tool call names the connection it targets. With no connections at all, only the documentation search and config-diagnostic tools are on. Tools can be narrowed further with --allow-tools and --block-tools.

The README talks of "50+ tools"; the tool-name list in the v1.6.0 source holds 73, including get-partition-offsets, which the README's tables leave out. Confluent also runs a fully managed MCP server for Confluent Cloud and points Cloud customers to it; this open-source one is the choice for Confluent Platform or self-managed Kafka, or for changing the toolset.

What it needs

Runtime	Node.js 22+

Node.js 22.19.0 or later (npx). A config.yaml, made with --init-config or --init-oauth-config, pointing at a reachable Kafka and/or Schema Registry, or a Confluent Cloud account with API keys or a browser sign-in. Cloud-only tools (Flink, connectors, Tableflow, metrics, billing) need Confluent Cloud. Tableflow tools also need IAM roles on your cloud account linked into Confluent Cloud first. pnpm only to build from source.

Limits

In v1.6.0, OAuth sign-in covers only the Kafka topic and message tools, topic config, Schema Registry, organization/environment/cluster listing and billing; Flink, connectors, Tableflow, metrics and catalog tags still need a direct connection with API keys (CONFIGURATION.md at v1.6.0; the main-branch README marks more as OAuth-ready, which is not yet released). On Confluent Platform or plain Kafka only the Kafka and Schema Registry tools work. Configuring by environment variables alone is deprecated and set to be removed. Confluent calls it community-supported: best-effort help, no SLA.

Who it suits

Suits teams on Confluent Platform or self-managed Kafka who want an assistant to inspect topics, consumer lag and schemas, and Confluent Cloud users who want Flink SQL, connector and Tableflow work from a local process with their own keys. Many tools write and delete (topics, schemas, connectors, Flink statements), so pair it with --allow-tools or a read-only key when the cluster matters. Confluent Cloud customers who want nothing to install can use Confluent's managed MCP server instead, which its README recommends for them.

Tools 60 of 73

list-topics: List the topics on a Kafka cluster
create-topics: Create one or more topics, with optional partition count and replication factor
delete-topics: Delete topics by name
produce-message: Write records to a topic, optionally serialized through Schema Registry
consume-messages: Read records from topics, from an offset, timestamp, earliest or latest
get-partition-offsets: Show low and high watermarks and message counts per partition of a topic
list-consumer-groups: List the consumer groups on a cluster
describe-consumer-group: Show a consumer group's state, protocol and member assignments
get-consumer-group-lag: Compute committed offset and lag per partition for one consumer group
get-topic-config: Read a topic's configuration
alter-topic-config: Change a topic's configuration on Confluent Cloud
create-flink-statement: Submit a Flink SQL statement
list-flink-statements: List Flink SQL statements with filtering and paging
get-flink-statement-results: Fetch the rows a Flink SQL statement produced
delete-flink-statements: Delete a Flink SQL statement
get-flink-statement-exceptions: Return the latest exceptions raised by a Flink statement
check-flink-statement-health: Rate a Flink statement as healthy, warning or critical
detect-flink-statement-issues: Look for problems in a Flink statement from its status, errors and metrics
get-flink-statement-profile: Return query profiler data and the task graph for a Flink statement
list-compute-pools: List the Flink compute pools in an environment
list-flink-catalogs: List Flink catalogs
list-flink-tables: List the tables in a Flink database
describe-flink-table: Show column names and types of a Flink table
list-connectors: List the active Kafka Connect connectors
get-connector-config: Read a connector's full configuration
get-connector-status: Show a connector's state and its tasks, with failure traces
get-connector-error-summary: Summarize a connector's current errors
get-connector-error-recommendations: Suggest fixes for a failed connector
get-connector-logs: Fetch recent connector log entries, the last hour of errors by default
create-connector: Create a new connector
update-connector-config: Replace a connector's configuration in full
pause-connector: Pause a connector and its tasks
restart-connector: Restart a connector and its tasks
delete-connector: Delete a connector
list-schemas: List the schemas in Schema Registry
create-schema: Register a schema or a new version under a subject
delete-schema: Delete a subject or one version of it
list-tags: List tag definitions in the catalog
create-topic-tags: Define new tags
add-tags-to-topic: Attach existing tags to topics
delete-tag: Delete a tag definition
search-topics-by-tag: Find topics carrying a given tag
search-topics-by-name: Find topics whose name matches a search
list-organizations: List the Confluent Cloud organizations the credentials can see
list-environments: List Confluent Cloud environments
list-clusters: List the clusters in an environment
list-tableflow-topics: List topics that have Tableflow enabled
create-tableflow-topic: Turn on Tableflow for a topic
delete-tableflow-topic: Turn off Tableflow for a topic
list-tableflow-regions: List the regions where Tableflow runs
create-tableflow-catalog-integration: Link Tableflow to an outside catalog such as AWS Glue
list-tableflow-catalog-integrations: List Tableflow catalog integrations
list-available-metrics: List the Confluent Cloud metrics and filter fields that can be queried
query-metrics: Query Confluent Cloud operational metrics
list-billing-costs: Return billing costs for an organization over a date range
search-product-docs: Search Confluent documentation by keyword
get-product-doc-page: Fetch a Confluent documentation page as markdown
explain-disabled-tools: Say why a tool is not enabled under the current config
list-configured-connections: List configured connections and the tools each one enables
config-help: Suggest the YAML needed to turn on a given tool

Popularity

63 popularity score, on our list (the bar is 36)

★ 169 ↓ 1.9k/wk

GitHub stars	169	26.8 points
Stars in 30 days	+2	2.4 points
npm downloads, last week	1,927	26.3 points
Official server	Yes	8 points

Measured 2026-11-07. How the score works.

Where it is published

Code	confluentinc/mcp-confluent ★ 169
npm	@confluentinc/mcp-confluent 1.6.1, 2026-10-26

What it can reach

Runs	On your machine
Reaches	The internet, Cloud resources, A database
Changes things	Yes: it can write, send, create or delete
Sign-in	An API key or token
Licence	MIT (npm)
Last code change	2026-10-26

Runs on your machine and connects out to the Kafka brokers, Schema Registry and Confluent Cloud APIs named in config.yaml, using the API keys there or, for an oauth connection, a Confluent Cloud browser sign-in reused for the life of the process. Tools can create and delete topics, write records, delete schemas, change and delete connectors, and run Flink SQL, all with the rights of those credentials. By default it sends usage data and error reports (Sentry) to Confluent; DO_NOT_TRACK=true turns both off. Over stdio there is no listener; with the http or sse transports it binds to 127.0.0.1, requires an API key in the cflt-mcp-api-key header and checks the Host header, unless --disable-auth is given.

Over time

GitHub stars 169 +2 (+1 %)

169167

11 Oct 20267 Nov 2026

npm downloads a week 1.9k +1.9k (+4280 %)

2.0k8

20 Oct 20257 Nov 2026

Before the site measured it, read from npm's daily download counts.

Popularity score 63 +11 (+22 %)

6352

11 Oct 20267 Nov 2026

What is rising across the directory

Sources

Page checked 2026-10-15.

Badge for your README

Shows this server's rank in its category, and stars for a server that runs on your machine. It updates every hour.

MarkdownREADME.md

[![Confluent MCP Server on EveryMCPServer](https://everymcpserver.com/badge/mcp-confluent.svg)](https://everymcpserver.com/servers/mcp-confluent)

Connect your client

Add in one click

Or add it by command or config

Add in one click

Or add it by command or config

Settings 9

What it does

What it needs

Limits

Who it suits

Tools 60 of 73

Similar servers

Confluent Cloud managed MCP server

StackQL MCP Server

Azure MCP Server

Render MCP Server

MCP DigitalOcean Integration

InfluxDB 3 MCP Server