Skip to main content
The StreamNative MCP Server lets AI agents and IDE copilots interact with your StreamNative Cloud clusters through the Model Context Protocol (MCP). MCP is an open standard that provides a unified way for AI models to discover and call external tools. With the StreamNative MCP Server, your agents can list topics, produce and consume messages, manage schemas, inspect cluster health, and more — all through natural language prompts in your IDE or agent runtime. StreamNative offers two deployment options so you can choose the setup that fits your workflow.

Deployment options

Local MCP Server

The local MCP Server is an open-source binary (snmcp) that runs on your machine. It connects to StreamNative Cloud, standalone Kafka clusters, or standalone Pulsar clusters. Best for:
  • Individual developers who want full control over their MCP setup
  • Connecting to external Kafka or Pulsar clusters outside StreamNative Cloud
  • Air-gapped or restricted environments where outbound connections to managed services are limited
  • Quick experimentation and prototyping

Remote MCP Server

The Remote MCP Server is a managed service hosted at https://mcp.streamnative.cloud. StreamNative operates the server so you do not need to install or update anything locally. Best for:
  • Teams that want a shared, always-current MCP endpoint with no local setup
  • Organizations that require centralized governance and access controls
  • Production environments where administrators control which tools and clusters are available

Comparison

FeatureLocal MCP ServerRemote MCP Server
InstallationInstall binary locallyNo installation required
UpdatesManual (Homebrew, Docker, or from source)Automatic — always current
StreamNative Cloud clustersYesYes
External Kafka clustersYesNo
External Pulsar clustersYesNo
Cluster-level governanceNoYes — access modes and allowed tools
AuthenticationService account API keysOAuth 2.1 or service account API keys
Multi-user sharingSelf-managed sharing (for example, SSE deployment with your own access controls)Shared endpoint across teams
Agents as toolsNoYes — Agents as Tools
Functions as toolsYes (Pulsar Functions)Yes — Functions as Tools

Get started

Learn more