Browse Workshops
Step 3

Create an MCP Bridge Instance

⏱ 4 min

Overview

In this step you’ll create the MCP Bridge instance. This is the gateway-side object that:

  • Listens for MCP protocol requests from AI agents
  • Routes each tool call to the correct API instance
  • Returns the HTTP response back to the agent in MCP format

Step 1 - Two ways to start

Option A — From an existing API

  1. Go to APIs and open the Photo Catalog API instance you created earlier
  2. Click Expose as MCP Server (top-right)
  3. Select APIs → Photo Catalog API

This selects the API for you and drops you into the bridge creation flow.

Option B — From the MCP Servers Menu

  1. Go to MCP Servers → Add MCP Server → Create MCP Server
  2. Select APIs → Photo Catalog API

Both paths lead to the same configuration screens. Use Option A for this workshop.

Troubleshooting — “No deployed instance available”

If you see the error “No deployed instance is available for this version. Pick a different version or deploy an instance before continuing.”, the API instance is missing a Consumer Endpoint. Fix it before proceeding:

  1. Switch to the legacy UI by clicking Anypoint Platform in the lower-left corner
  2. Go to APIs → API Manager
  3. Find and click Photo Catalog API
  4. Click Edit Instance Configuration
  5. In the Consumer Endpoint field, enter your Omni Gateway base URL with the API path appended: 
    <your-omni-gateway-base-uri>/photo-catalog

    For example: https://attendee-gateway-abc123.cloudhub.io/photo-catalog

  6. Click Save
  7. Return to the new UI and retry creating the MCP Server

Step 2 — Select APIs and tools

On the APIs step:

  1. The Photo Catalog API is pre-selected (checked) if you used Option A. You can add more APIs — each becomes a group of tools in the bridge.

  2. From Select Tools, choose both the resources from API. Notice that you can change the tool name, and description if required.

    Why tool definitions matter

    An AI agent doesn’t read your API spec at runtime. Instead, it reads the tool name and description you provide here to decide:

    • Whether a tool is relevant to the user’s request
    • What parameters to pass
    • How to interpret the response

    Writing clear, specific descriptions is the single most impactful thing you can do for agent quality. A vague description like "Gets photos" will be ignored or misused. A precise one like "Search the Photo Catalog by category. Use this when the user wants to browse or filter available photos." tells the agent exactly when and how to use it.

    The auto-generated names work, but more descriptive names and richer descriptions improve agent behavior.

    Tips for writing good tool descriptions

    Do

    • State the purpose — what task this tool accomplishes
    • List when to use it — the user intent that should trigger this tool
    • Mention dependencies — e.g. “requires an ID from list_photos”
    • Describe filters or options available

    Don’t

    • Use technical jargon the agent won’t understand (“invokes the GET /photos endpoint”)
    • Be too brief (“gets photos”)
    • Duplicate descriptions across tools

    Example: weak vs. strong description

      Description
    Weak Get photo by ID
    Strong Retrieve the price, description, and stock level for a specific photo. Use this when the user asks about a photo they found from list_photos. Requires the exact photo ID.

  3. Click Next

  4. Provide description of this MCP server

    This MCP Server provides tools to retrieve all the photos (listphotos) as well as  price, description, and stock level for a specific photo (getphoto). Use this tool when the user asks about a photo they found from list_photos. Requires the exact photo ID

  5. Click Create & Deploy

When everything looks correct, click Create & Deploy.

What’s next

The MCP Bridge is now deployed. The next step is testing the MCP Bridge

Continue to Test Your MCP Bridge →