StreamsHub MCP

StreamsHub MCP provides Model Context Protocol (MCP) servers that give AI assistants direct access to Kubernetes-based streaming infrastructure.
Each component of your streaming stack gets its own MCP server that can be deployed independently.

Works with any MCP-compatible client including Claude Code, Claude Desktop, VS Code, Copilot, and more.

MCP servers

Key capabilities

StreamsHub MCP servers leverage the full MCP specification:

• Prompt templates -- Structured diagnostic workflows that guide the LLM step-by-step through complex troubleshooting
• Resource templates -- Live cluster state exposed as structured context that clients can attach directly to conversations
• Resource subscriptions -- Kubernetes watches that push real-time notifications when cluster state changes
• Completions -- Dynamic autocomplete for parameters (namespaces, cluster names, topics) powered by live Kubernetes queries
• Progress and cancellation -- Long-running operations report progress and support mid-operation cancellation

Getting started

Pick a server from the table above and follow its getting started guide.
For the full documentation, see the documentation/ directory (modular AsciiDoc).
Build it to HTML locally with ./mvnw -N -Pdocs package (output in target/documentation/).

Development

./mvnw compile              # Compile all modules (includes checkstyle)
./mvnw test                 # Run unit tests
./mvnw verify -Psystemtest  # Run system tests (requires Kubernetes cluster)

For more details:

• Contributing -- Contribution workflow and guidelines
• Testing -- Unit and system test guide

Built with

• Quarkus -- Cloud-native Java framework
• Fabric8 Kubernetes Client -- Kubernetes API access
• Quarkus MCP Server -- MCP protocol implementation

License

Apache License 2.0