Skip to content
MCP server

Tapp

Lets coding agents verify UI changes on iOS, Android and web apps and turn them into CI

Runs locallyStatus and history
Install
Pricing
Open source
Platforms
macOS, CLI, API
Runs
Locally
Version
0.35.7
Listing updated
today

Similar MCP servers

About Tapp

Tapp is an open-source MCP server for developer tools. It runs locally on your machine and is started by your AI client.

The README says Tapp gives coding agents a way to look at and drive real iOS simulator, Android emulator or device, and browser screens, so UI edits can be checked against the running app rather than assumed. It can also explore an app for technical failures such as crashes, dead buttons and stuck loading screens, and report findings with coverage and screenshots.

The project separates observation from judgment. Exploration produces findings and limits, while a deterministic gate returns pass, fail or inconclusive based on committed Flows, reviewed contracts and a baseline. Journeys saved during a session can be replayed as deterministic tests, and a GitHub Action can run the same checks on pull requests.

It suits teams with their own iOS, Android or web apps who want agent-driven verification and regression checks in CI. Exploration is meant for apps the user owns, with a read-only mode for sites they do not.

Install

Run in a terminal

claude mcp add tapp -- npx -y @aarwitz/tapp

Add to claude_desktop_config.json

{
  "mcpServers": {
    "tapp": {
      "command": "npx",
      "args": [
        "-y",
        "@aarwitz/tapp"
      ]
    }
  }
}
Install in CursorOpens Cursor, which asks you to confirm. Or add it by hand:

Add to ~/.cursor/mcp.json

{
  "mcpServers": {
    "tapp": {
      "command": "npx",
      "args": [
        "-y",
        "@aarwitz/tapp"
      ]
    }
  }
}
Install in VS CodeOpens VS Code, which asks you to confirm. Or add it by hand:

Add to .vscode/mcp.json

{
  "servers": {
    "tapp": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "@aarwitz/tapp"
      ]
    }
  }
}

Add to ~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "tapp": {
      "command": "npx",
      "args": [
        "-y",
        "@aarwitz/tapp"
      ]
    }
  }
}

Add to your client's MCP settings (most use this shape)

{
  "mcpServers": {
    "tapp": {
      "command": "npx",
      "args": [
        "-y",
        "@aarwitz/tapp"
      ]
    }
  }
}

Using several servers? Build one config file with the MCP config generator.

Tools this server provides (13)

Taken from the project documentation; the running server may differ.

tapp_look
See a screen: current surface summary with inline screenshot; optional accessibility tree.
tapp_session
Interactive driving with tap, type, swipe, back, wait and login actions, with optional save as Flow.
tapp_explore
Autonomous exploration returning an observation of findings, without a verdict.
tapp_init
Detect targets in a repository and persist the shared UI Map and release plan.
tapp_plan
Release-plan lifecycle including show, review, generate, validate and promote.
tapp_run
Deterministic replay of a committed Flow, Task, contract or Scenario.
tapp_validate
Validate a committed Flow, Task, contract or Scenario file.
tapp_ci
Save a target-scoped baseline or install the CI workflow.
tapp_map
Build, inspect, merge and diff the persistent UI Map.
tapp_build
Build, install, list and boot iOS simulator apps.
tapp_doctor
Check the local toolchain.
tapp_report
Show capture history and evidence for the project.
tapp_feedback
Draft or submit an issue about Tapp itself.

What it does

Agent screen inspection
Shows the current screen with a summary, screenshot and optional accessibility tree for an iOS, Android or web surface.
Autonomous exploration
Explores an app without a scripted test and reports findings such as crashes, failed sign-ins and broken links.
Replayable flows
Saves driven journeys as committed Flows that replay deterministically without an AI model at runtime.
Deterministic CI gate
Returns pass, fail or inconclusive from contracts, Flows and baseline regressions, and can run on pull requests.
Read-only mode
Explores live or third-party sites without clicking, typing or submitting.

Good for

  • Have a coding agent confirm a UI change renders correctly on a simulator before reporting it done
  • Find crashes and dead ends in an app before a release
  • Turn a manually driven login-and-checkout journey into a repeatable regression test
  • Gate pull requests on new high-severity findings or broken journeys in GitHub Actions

Pricing

Open source

MIT-licensed core; optional paid founding pilot for teams

Tapp FAQ

Does Tapp need an API key or account to run?

The README says core exploration, Flow replay, and CI gating require no API key and no coding agent at runtime; AI features are optional and explicit.

Which platforms are supported?

iOS simulators on macOS with Xcode, Android emulators or devices via adb, web apps in Playwright Chromium, and a macOS desktop target, with Windows desktop support in early pilot runs.

What is the license?

The README states that Tapp's core is MIT licensed.

Can it test a site I do not own?

Use the read-only exploration option, which renders and reads the page without clicking or submitting. The README says exploration is intended for apps and environments you own.

Is Tapp free?

Yes. Tapp is open source, so you can use and self-host it for free.

How do I add Tapp to Claude Code?

Run this command in your terminal: claude mcp add tapp -- npx -y @aarwitz/tapp

What are the best alternatives to Tapp?

The closest alternatives to Tapp are looktap, Tapir, CheckMyApp, Tavero and Swipium.

Where this comes from

  • Description: written from the project's README and its MCP Registry entry.
  • Install steps: from the official MCP Registry entry.

How AIPoweredBot checks servers