User Guides
MCP
Playground

MCP Playground


The MCP Playground is a visual interface for inspecting, testing, and debugging MCP servers. It allows developers to simulate requests, inspect tool schemas, and validate execution results — all without writing client code.

This guides walks through:

  • What the Playground is and why it's useful
  • Core features like tool inspection and manual testing
  • Common workflows (e.g., testing STDIO vs. SSE)
  • How to troubleshoot errors before production deployment

Introduction

The Playground is a developer tool that connects directly to your MCP server and gives you a live interface to:

FeatureDescription
Connect to a serverChoose STDIO or SSE, provide the server path or URL
List toolsAutomatically fetch tool schema from capabilities.json
Test tool callsManually input parameters and run the tool
View responsesSee formatted output, errors, or return values
Track tool historyReview previous executions in a live session

It's your "debug console" for validating server behavior before clients call it. You can also access it via modelcontextprotocol.io/inspector (opens in a new tab).

Debugging

The Playground saves hours by letting you: Debug tool behaviors quickly; Share visual results with your team; Confirm server compatibility before public release.

Depending on the connection type, the Playground works slightly differently:

ConnectionSetup NeededNotes
STDIOLocal path to .js or .py fileServer is launched in background
SSEURL to a cloud endpoint (e.g., Vercel API route)Must support streaming and return valid tool list

Once connected, the Playground:

  • Sends a list_tools request
  • Shows available tools with their input schema
  • Lets you fill in test values and submit a request

Use Case

To verify the MCP server:

  1. Click Playground from the header menu.

playground

  1. In the MCP Inspector section:

    • Choose the MCP Server type: STDIO or SSE
    • Enter the service URL and click Connect
    • If successful, the status will turn green and show Connected
  2. Under the Tools section, click List Tools to view all available tools defined in your API.

run-tool

  1. Select and run individual tools by clicking Run Tool
  2. Input the required parameters.
  3. View the execution response in the History section.