Model Context Protocol

Test APIs with
AI assistance

The API Unit MCP server connects your AI assistant directly to your testing platform. Generate test cases and run them — in natural language, inside any MCP-compatible client.

What is MCP?

Model Context Protocol (MCP) is an open standard that lets AI assistants call external tools and services. Instead of copy-pasting data back and forth, your AI can directly invoke actions — like querying a database, searching docs, or in this case, generating and running API tests.

Think of it as a plugin system for AI: you add the API Unit MCP server to your client once, and from that point on your assistant can interact with your testing platform as part of any conversation.

AI client

Claude Desktop, Cursor, …

MCP protocol

Standardised tool interface

API Unit

Generate & run tests

See it in action

A real conversation using the two available MCP tools.

API Unit MCP — Claude Desktop
MCP connected

Press Play demo to see the MCP tools in action

Simulated conversation

Click the tool call rows to expand the raw JSON.

Available tools

Two focused tools — one to create, one to execute.

generate_test_case

Create and persist a test case in ApiUnit. Returns the stored test case with its ID.

Parameters

  • · name (string, required)
  • · http_status (number, required)
  • · url (string)
  • · endpoint_id (string)
  • · method (string)
  • · body_parameters (object)
  • · query_parameters (object)
  • · headers_request (object)
  • · headers (object)
  • · load_speed (number)
run_test_case

Execute an existing test case by ID and return the live response.

Parameters

  • · test_case_id (string, required)

Setup in 3 steps

Get connected in under two minutes.

1

Generate an API key

Log in to app.apiunit.io, go to the API Keys page, and generate a new key.

2

Add the server to your MCP client

Paste the snippet below into your Claude Desktop claude_desktop_config.json (or equivalent config file for Cursor / other clients).

json — claude_desktop_config.json
{
  "mcpServers": {
    "api-unit": {
      "command": "https://apiunit.io/api-unit-mcp",
      "env": {
        "API_UNIT_URL": "https://api.apiunit.io",
        "API_UNIT_TOKEN": "your_token_here"
      }
    }
  }
}

Replace your_token_here with the token you copied in step 1.

3

Restart and start testing

Restart your AI client. The API Unit tools will appear automatically. Try asking: "Generate a test for my GET /products endpoint and run it."

Why use the MCP integration?

Natural language interface

Describe what you want to test in plain English. No need to remember endpoint schemas or assertion syntax.

Generate & run in one step

Ask an AI to create a test case and execute it immediately — all without leaving your conversation.

Plug into any MCP client

Works with Claude Desktop, Cursor, and any editor or tool that supports the Model Context Protocol.

Secure by design

Your token stays in your local config. The MCP server never stores credentials.

Ready to test smarter?

Create a free account, grab your token, and connect in minutes.

Create free account