While building a UI-first application, I kept running into the same problem: my AI agent was generating mock data with static strings and weak examples that did not feel realistic enough for real product work. That frustration led me to build JsonPlace.

JsonPlace MCP is an tool that combines Faker-style field generation with real remote mock endpoints so agents can generate better payloads and actually serve them during development. Another big advantage is that creation is not LLM-based, which saves context, reduces token usage, and makes mock data generation more deterministic.

This is the first public version of the idea. It is completely free and open source, and I would genuinely love to hear feedback, ideas, and real use cases from other developers.