MCP SERVERS

Mock MCP

Expose mock server routes as MCP tools for AI clients — predictable, controlled API responses without hitting real endpoints.

Overview

Enable MCP on a mock server to expose its routes as MCP tools at /mcp. When an AI client calls a tool, it receives the canned mock response body.

Useful for giving AI assistants access to predictable, controlled API responses without hitting real endpoints.

Mock server settings → toggle 'MCP' on (mcpEnabled: true).

Adds a /mcp endpoint alongside regular mock routes

Same port as the mock server

MCP endpoint at: http://localhost:{port}/mcp

Each mock route with an mcpDescription becomes an MCP tool.

Route without mcpDescription: not exposed as a tool (still works as a normal mock route)

Tool name derived from method + path (e.g. GET_users)

Tool call returns the configured mock response body

In the Route Editor, expand the 'MCP Tool' section.

Field	Type	Description
mcpDescription	string	What the tool does — shown to AI clients
mcpInputSchema	JSON Schema	What parameters the tool accepts

Both fields are optional. Without a description, the route is hidden from the MCP tool list.

AI client calls tool → mock server returns the configured response body.

No actual HTTP request is made to a real server

Response is instant (no network latency)

Status code from route config included in response metadata

Great for testing AI integrations without backend dependencies

Give them mock API access during development

Verify AI correctly interprets schemas and parameters

Showcase AI integration without live APIs

AI always gets consistent data for testing logic

Management MCP → Connecting AI Clients →