Skip to content

Understanding MCP and Wanaku

Welcome to the Wanaku demo series. Before writing any commands, this guide explains the concepts behind the Model Context Protocol (MCP) and how Wanaku fits into the picture.

TIP

Prefer video? Watch MCP in Action: Connecting AI to Enterprise Systems by Zineb Bendhiba for a walkthrough of MCP in practice.

What You Will Learn

  • What MCP is and why it exists
  • The three MCP primitives: tools, resources, and prompts
  • How Wanaku extends MCP with routing, namespaces, and forwarding
  • When to use Wanaku instead of a single MCP server

No Setup Required

This is a conceptual guide. You do not need to install anything or run any commands. Read through it before moving on to Demo 1.01: Getting Started.

Step 1: What Is MCP?

The Model Context Protocol (MCP) is an open protocol that standardises how AI agents communicate with external tools and data sources. Think of it as a common language that lets an AI agent ask:

  • "Run this tool and give me the result" — a tool call
  • "Fetch this piece of data for me" — a resource read
  • "Give me a template I can fill in" — a prompt

Before MCP existed, every AI application had to write its own integration layer. MCP removes that duplication by defining a single, reusable protocol.

Step 2: The Three MCP Primitives

PrimitivePurposeExample
ToolAn action the agent can invokeCall a REST API, run a shell script
ResourceA data source the agent can readA file, a database row, an S3 object
PromptA reusable text templateA system prompt, a query template

A single MCP server exposes one or more of these primitives. An AI agent connects to an MCP server and discovers what tools, resources, and prompts are available.

Step 3: The Problem Wanaku Solves

A single MCP server works well for small use cases, but real deployments face three challenges:

Multiple capability services. You might have an HTTP capability, a database capability, and a Kafka capability each running as a separate service. Without a router, the agent would need to know about all of them.

Namespace isolation. In a team environment, the dev namespace and the production namespace should see different tools. Hardcoding that logic in every capability service is brittle.

Forwarding. An enterprise deployment might want to aggregate tools from several upstream MCP servers into a single endpoint. Without forwarding support, agents have to manage multiple connections.

Step 4: How Wanaku Solves It

Wanaku acts as an MCP router that sits between your AI agent and your capability services.

The agent connects to one endpoint, the Wanaku Router. The router maintains a registry of capability services and forwards requests to the right one. Namespaces let you partition tools and resources without running separate routers. MCP Forwards let Wanaku aggregate tools from upstream MCP servers transparently.

Step 5: When to Use Wanaku

ScenarioRecommendation
One agent, one toolA single MCP server is enough
Multiple capability types (HTTP, exec, database)Use Wanaku to route between them
Team with dev/staging/production isolationUse Wanaku namespaces
Aggregating upstream MCP serversUse Wanaku MCP Forwards
Kubernetes/OpenShift deploymentUse the Wanaku Operator

What's Next?

Now that you understand the concepts, move on to the hands-on guides:

If you find a bug, please report it. To get in touch with the community, visit the Wanaku project.