mcpdirectory.dev
9,129 servers · updated 23 Sept

Mockserver MCP server

official from Mock Server Active last commit 18 d
{
  "mcpServers": {
    "mockserver": {
      "command": "docker",
      "args": [
        "run",
        "-i",
        "--rm",
        "docker.io/mockserver/mockserver:7.6.0"
      ]
    }
  }
}
file .mcp.json (in your project root)version 8.0.0

What it does

MockServer is an HTTP(S) mock server and proxy for testing that mocks APIs, records and inspects traffic, and injects failures for chaos testing. It supports HTTP/1.1, HTTPS, HTTP/2, HTTP/3, gRPC, WebSockets, JSON-RPC, and message brokers, and includes mocking for LLM/AI chat-completion APIs. It is for developers and QA engineers who need to test applications against unavailable, incomplete, or unreliable dependencies.

  • Mock external APIs during development when those services are unavailable or incomplete, then verify your application calls them correctly.
  • Record and replay HTTP traffic in a proxy to inspect requests and responses, set breakpoints, and modify exchanges in flight like a network debugger.
  • Inject latency, dropped connections, and errors into dependencies to test how your application handles degraded or failing services.

Tools this server exposes

ToolDoes
mockserver_create_expectationCreate a mock expectation that matches HTTP requests and returns a configured response.
mockserver_verify_requestsAssert which requests were received, in what order, and how many times.
mockserver_inject_failureInject latency, dropped connections, or error responses into a mock to test resilience.
mockserver_record_trafficProxy and record real traffic, with the ability to inspect and modify requests and responses.
mockserver_mock_from_openapiGenerate mock expectations from an OpenAPI or Swagger specification.
mockserver_mock_llm_apiMock chat-completion APIs for OpenAI, Anthropic, Gemini, Bedrock, Azure OpenAI, and Ollama, including streaming responses.

Before you enable it

An MCP server runs with whatever access you give it. Start with a development account rather than production, and check which tools the server exposes before you let an agent call them unattended.