Pour yourself a
mock API.

A complete mock server and dashboard in one little binary — mixed fresh on your machine. No cloud, no sign-up, no bitter aftertaste.

Start mixing → Star on GitHub
brew install --cask mocktail — a native app, on your desktop
The whole thing

A dashboard that comes in the box.

Freshly mixed, every request.

Model the responses your app expects — headers, delays, even the flaky 401 — and let Mocktail keep the data lively.

Randomize · optional

Fixed by default, fresh on demand

Serve a plain static response as-is — or attach a generator to any field (uuid, email, price, name) so it re-rolls every call. Opt in per field; point one at a real response and it anonymizes it.

GET/api/v1/users · 200
{
  "id":    "3f9a8c1e"     ⟳ uuid
  "name":  "Ada Lovelace" ⟳ fullName
  "email": "ada@hey.dev"  ⟳ email
}
Live traffic

Watch every request land

A real-time stream of everything hitting your mocks — method, status, latency, and the exact response and headers served. Filter by path or status, replay at a glance.

live
GET/orders12 ms
POST/users8 ms
GET/products3 ms
POST/auth/login401 · 5 ms
Headers & latency

Model the messy parts

Real APIs aren't always a happy 200. Return a 429 with Retry-After, add 800 ms of latency, serve a problem+json error — whatever your app has to handle.

response headers
HTTP/1.1 429 Too Many Requests
Retry-After:   30
Content-Type:  application/problem+json
X-RateLimit:   0/100
AI assistant

Describe it, don't type it

Bring your own key. Say what you want in plain language and Mocktail drafts the whole response with Claude, right in the dashboard. Your key never leaves your machine; chats are never stored.

🔑 BYO key
✳ assistant
✳ an order with line items and a total
drafts →
{
  "id": "ord_8a2f",
  "items": 3,
  "total": "$84.20"
}
MCP

Or let your AI tools drive

The built-in MCP server plugs Mocktail into Claude Desktop & Code. Ask, and they create and edit mocks for you — a sentence becomes a live endpoint, no dashboard required.

MCP server
claude · mcp
youadd a paginated /orders endpoint
claude✓ created GET /api/v1/orders → 200

Your first pour.

The dashboard comes with it. Open localhost:6625 and start mocking.

$
brew install --cask mocktail
# a native app — auto-picks a free port

Questions, answered.

What is Mocktail?

Mocktail is a self-hosted mock API server with a built-in dashboard, shipped as a single binary. You define endpoints and their responses, and your app calls them like a real API — locally, with no cloud service.

Is Mocktail free and open source?

Yes. Mocktail is free and open source on GitHub. There's no sign-up and no paid tier to use it — it runs entirely on your own machine.

How do I install Mocktail?

Install the desktop app with Homebrew (brew install --cask mocktail), run it with Docker, or use the CLI. The dashboard opens at localhost:6625.

Where does my data live?

Everything runs locally and is stored on your machine — self-hosted, no cloud, no telemetry. Your mocks are plain JSON you can export, commit, and share.

Can Mocktail generate realistic test data?

Optionally. Serve a fixed response as-is, or attach a generator to any field (uuid, email, price, name) so it returns fresh data on every request. Point one at a real response and it anonymizes it.

Does Mocktail work with AI tools?

Yes. A built-in AI assistant drafts responses from a plain-language prompt (bring your own key, key stays on your machine), and the MCP server lets Claude Desktop and Claude Code create and edit your mocks.