MCP Server Configuration with Pcloudy

Prev Next

Overview

Pcloudy now supports MCP (Model Context Protocol) Server, allowing users to connect to real mobile devices from their own AI tools, IDEs, or custom automation frameworks. This integration makes it possible to enable autonomous workflows, where AI agents or external systems can interact with devices hosted on Pcloudy in real time. Whether you're building intelligent test orchestration, running autonomous test suites, or developing custom automation logic, MCP Server gives you the flexibility to programmatically control devices without relying on manual intervention or UI-based interactions. It’s a significant step forward in enabling agentic AI-driven testing and seamless device access from remote environments. With secure and scalable protocol-level connectivity, teams can now fully integrate Pcloudy devices into their intelligent DevOps pipelines and next-gen testing infrastructure.

With this support, users can:

  • Establish direct control over real devices hosted on Pcloudy

  • Enable AI agents to interact with mobile devices in real time

  • Integrate Pcloudy devices into autonomous testing workflows

  • Connect from custom environments, IDEs, or orchestration platforms

Use Case Examples:

  • Autonomous app testing using AI agents

  • Custom device interaction workflows driven by external orchestration systems

  • Continuous testing pipelines with real device access

What is MCP?

MCP (Model Context Protocol) is an open standard that simplifies how applications interact with large language models (LLMs). It's like a USB-C port for AI: a single, consistent way to connect tools and data sources to your LLMs.

MCP allows you to:

  • Plug LLMs into existing tools and APIs with ease

  • Switch between different LLM vendors with minimal effort

  • Keep data secure by executing within your infrastructure

  • Build composable, modular workflows for real-world use cases

Learn more about MCP

About Pcloudy-MCP

Pcloudy-MCP serves as an MCP server that interacts with the Pcloudy Device Cloud to:

  • Manage test cases via natural language

  • Execute manual or scripted tests on Android/iOS devices

  • Automate test steps using Qpilot, an LLM-compatible AI testing agent

Why Use It?

  • Test from Anywhere - Easily control real devices using natural prompts — whether you're in a terminal, IDE, Claude, Cursor, or any other AI tool.

  • Reduce Context Switching - Trigger device actions, upload builds, or run flows directly from your development or chat environment — no more jumping between tools.

How to Use

Prerequisites

  1. Sign up for Pcloudy account → https://device.pcloudy.com/signup

  2. Install Python ≥ 3.10 on your machine

  3. Install uv: https://docs.astral.sh/uv/getting-started/installation/#standalone-installer

Configuration (MCP-compatible tools)

To launch the Pcloudy MCP server, use this configuration:

  1. For Claude claude_desktop_config.json

{
"mcpServers": {
"Pcloudy": {
"command": "uvx",
"args": ["pcloudy-mcp"],
"env":{ 
"PCLOUDY_USERNAME": "<your_pcloudy_username>", 
"PCLOUDY_API_KEY": "<your_pcloudy_api_key>", 
"PCLOUDY_CLOUD_URL": "<your_pcloudy_cloud_url>"
} 
} 
} 
}
  1. For Cursor mcp.json

{
"mcpServers": { 
"Pcloudy": { 
"command": "uvx", 
"args": ["pcloudy-mcp"], 
"env": { 
"PCLOUDY_USERNAME": "<your_pcloudy_username>", 
"PCLOUDY_API_KEY": "<your_pcloudy_api_key>", 
"PCLOUDY_CLOUD_URL": "<your_pcloudy_cloud_url>"
} 
} 
} 
}

Use Case Examples

Manual Testing (Natural Language)

  • Upload the APK to the cloud

  • Install the app on a Pixel 6

  • Resign and upload the iOS build

Qpilot Automation

Book another iOS device and run the following:

  • Enter username 'user@test.com'

  • Enter password 'pass'

  • Click on login