Read the screen
Capture screenshots and streaming context so the agent can reason about the current phone state.
MCP integration
MCP iPhone automation for AI agents
TapKit connects real iPhones to MCP-compatible agents, giving them the tools to inspect the screen, choose an action, and operate the phone.
Protocol
MCP tools
Expose phone actions to agents through Model Context Protocol tools instead of custom glue code.
Device
Real iPhone
Run the agent loop against a physical iPhone with installed apps, accounts, notifications, and mobile-only UI.
Use
Agent clients
Connect TapKit to MCP-capable environments for experiments, demos, QA, and internal automation.
Why MCP
Give the agent phone tools where it already works
MCP is useful because the agent does not need a bespoke integration for every environment. TapKit can expose phone-control actions as tools the agent can call from an MCP-compatible client.
The workflow is straightforward: the agent sees the screen, chooses an action, TapKit executes it on the physical iPhone, and the agent observes the next state.
That makes MCP a strong starting point for mobile agent experiments, QA workflows, and internal tools that need to interact with real iOS apps.
Capabilities
What MCP agents can do with TapKit
Act on the phone
Tap, swipe, type, open apps, run shortcuts, use navigation controls, and move through mobile workflows.
Operate across apps
Let the agent move between Messages, Settings, social apps, files, web views, and other mobile surfaces.
Keep humans in control
Monitor sessions, interrupt work, and review what the agent saw and did before operationalizing a workflow.
Fit
When to use the TapKit MCP integration
- You want an MCP-compatible agent to use a real iPhone as a tool.
- You need mobile workflows that do not have complete public APIs.
- You are prototyping agentic phone workflows before building a full integration.
- You want the same control layer to work later through REST APIs or the Python SDK.
Path to production
Start with MCP, graduate when needed
MCP is a practical way to prove the workflow and teach agents how to use phone tools. When you need repeatability, service integration, or backend orchestration, the same TapKit phone layer can be used through the REST API or Python SDK.
That keeps the prototype close to the production path instead of turning a demo into a dead end.