Mcp

Mcp (Phone Harness) enables developers to create AI agents that can directly interact with real iPhones and Android devices—letting agents perform tasks like tapping, typing, and scrolling apps, just as a human would. The offering provides the ability to use either hosted devices in the cloud or to connect your own physical phones, with no jailbreaks, custom APIs, or additional apps needed. It is open-source and supported by documentation and community resources.

The AI agent "sees" the phone's screen using either OCR or accessibility data, which allows it to interpret what is shown and to operate the device interface accordingly. Persistent app states, real device phone numbers, and handling for SMS/2FA workflows are provided, making it practical for automated testing, demonstrations, or development use cases where interacting with actual mobile device interfaces (not just APIs or emulators) is required.

The platform is designed to scale, supporting everything from single device connections for development to large-scale phone farms for QA or automation purposes. Pricing starts with a pay-as-you-go option for Android emulators and a monthly flat-rate subscription for real iPhones, making experimentation and real-world integration accessible for both individuals and organizations. Multiple concurrency options and custom plans are available for dedicated phone workflows, suiting more intensive or production-grade needs.

Mcp is ideal for software developers, QA engineers, and technical product managers working on automation, testing, or demonstration of mobile applications. Teams building or evaluating AI agents that need to interact with real device interfaces—not just simulators or APIs—will benefit from its ability to operate physical or hosted phones at scale, enabling advanced automation and validation scenarios.

It is particularly relevant for startups exploring AI-driven automation of user tasks on mobile, companies needing to test cross-device workflows (including 2FA/SMS), and researchers or developers exploring agent-driven UI control. Both independent developers and engineering teams in mobile-focused organizations can use the tool for remote testing, device farms, or integrating real-device touch/interaction flows into their pipelines.