Tell Claude, Cursor, or any LLM to spin up a mock GraphQL API. They build it, you watch, and the link is in your clipboard before the model finishes its sentence. No login. No config files.
Whether you're typing a one-liner into ChatGPT or wiring up a long-running Cursor session, there's a path. They all hit the same public API and end with a queryable mock.
Tell any LLM to use Mockomat.
No tools to install, no API keys to copy. The capability spec at /.well-known/ai-capabilities.md teaches every LLM how to talk to Mockomat. The model spins up a guest project, builds your schema, and hands back a queryable URL.
"Create a mock GraphQL API for an e-commerce store with products, customers, and orders. Use Mockomat." Native integration for Claude Desktop, Claude Code, Cursor, Copilot.
mockomat-mcp exposes 10 tools — create_project, list_data_sources, create_table, generate_api, query_data, import_openapi, and more. The session cookie is managed for you. The AI never sees HTTP, just clean tool calls.
$ claude mcp add mockomat -- npx -y mockomat-mcp Have a Swagger or OpenAPI file? Mockomat reads it.
Attach a spec to your prompt and the LLM calls import_openapi. Tables, attributes, relations, and a full GraphQL surface — all generated from your existing API contract. Iterate on the model from there.
$ mockomat import petstore.openapi.yaml
→ 3 tables · 24 attributes · 12 operations Modelling decisions normally take days of meetings — the PM disagrees with the architect, the dev is in the dark, the tester is waiting for a contract. On Mockomat, the AI builds the schema in real time and your whole team watches the same board. Tables appear. Attributes bind. Relations connect. Everyone sees the same picture, at the same moment.
mockomat-mcp is a published npm package that speaks the Model Context Protocol over STDIO. Add it once to Claude Desktop, Claude Code, or Cursor — your AI gains a new toolbox.
claude mcp add mockomat -- npx -y mockomat-mcp{
"mcpServers": {
"mockomat": {
"command": "npx",
"args": ["-y", "mockomat-mcp"]
}
}
}The free entry points are wide enough for any prompt to succeed. Production workloads — CI integration, public demos, AI agents in the loop — live on a paid plan with API-key auth and higher quotas.