For AI agents

A safe API sandbox your AI agents can build on

Connect Cursor, Claude Code, Claude Desktop or ChatGPT to MockAPI over the Model Context Protocol. Agents design and create mock APIs from a prompt, and test against an isolated mock instead of production.

Start building free MCP setup guide

Free forever plan · No credit card required

~/.cursor/mcp.json

{
  "mcpServers": {
    "mockapi": {
      "url": "https://mcp.mockapi.com/api/mcp",
      "headers": {
        "Authorization": "Bearer mockapi_mcp_YOUR_KEY"
      }
    }
  }
}

Agents hitting real systems

Autonomous code that calls production or staging APIs is a risk.

Flaky agent runs

Upstream data that changes between runs makes agent loops hard to repeat.

Context switching

Leaving the editor to click through a dashboard breaks flow.

What you get

MCP tools for everything

list_projects, create_model, create_rest_endpoint, create_graphql_endpoint and more.

Import from specs

Agents can import OpenAPI schemas and Postman collections in one call.

Stored test records

enable_stateful_mode turns on stored records. The first GET to each list endpoint seeds them (25 by default, up to your plan limit), and agents then read and change them over HTTP like a real API (Developer plan and above).

Design guidance

A built-in API design guide helps agents follow REST conventions.

Scoped API keys

MCP keys use scopes, can be revoked anytime and never touch production data.

Works with any MCP client

Get started in three steps

01
Create an MCP key
Settings → API keys → create a key with the scopes your agent needs.
02
Add the server
Paste the config into Cursor, Claude Code or Claude Desktop.
03
Prompt
“Create a mock API for a todo app with users and tasks.”

FAQ

Common questions

What is MCP?

The Model Context Protocol is an open standard that lets AI assistants call external tools. MockAPI exposes its project, model and endpoint operations as MCP tools.

Which AI tools are supported?

Any MCP client, including Cursor, Claude Code, Claude Desktop and ChatGPT (where MCP connectors are available).

Is it safe to let agents create mocks?

Yes. Agents only reach your MockAPI projects through your MockAPI sign-in or a scoped key you can revoke, and mocks are isolated from your production systems.

Frontend development API testing & QA Mobile development OpenAPI mock server GraphQL mock server Free fake REST API Compare tools

Build against a mock API that behaves like your real one

Create your first endpoint in under two minutes.

Start building free View pricing

Free forever · No credit card required