Sauce MCP

Beta

Sauce MCP is currently in beta. Capabilities and configuration may change.

Sauce MCP lets your AI assistant or agent talk to Sauce Labs directly. It is a hosted Model Context Protocol (MCP) server that exposes Sauce Labs capabilities as tools any MCP-compatible AI client can call using natural language.

What is Sauce MCP?

The Model Context Protocol (MCP) is an open standard that lets AI applications connect to external tools and data sources. An MCP server publishes a set of tools; an MCP-compatible client (such as an AI coding assistant) can then call those tools on your behalf while answering a prompt.

Sauce MCP is a hosted MCP server operated by Sauce Labs. Once you connect a supported AI client, your agent can work with the Sauce Labs platform through natural language instead of switching to the dashboard or writing API calls by hand.

What you can do

With Sauce MCP connected, your AI agent can:

Discover available real devices and check their status
Start, inspect, and end real-device sessions
Install, launch, and uninstall apps from Sauce Storage
Capture screenshots, read the UI, and run shell commands on a device
Configure network conditions and capture network traffic
Inspect jobs, builds, and test artifacts
Look up account, team, and tunnel information
Generate test cases from natural-language intent and export them as runnable source code
Run test cases and suites on Sauce Labs and schedule them to run on a recurring cadence

See Tools for the full breakdown of capabilities, and the Prompting Guide for ready-to-use prompts. Test authoring capabilities are powered by Sauce AI for Test Authoring, a paid add-on for Enterprise accounts. Mobile device capabilities are currently limited only to Private Devices using Access API, a paid add-on to Real Device Cloud.

Who it's for

Sauce MCP is built for both technical and non-technical users. If you can describe what you want in plain language, you can drive a real device and author tests: no Appium scripts, no dashboard navigation, and no prior automation experience required. Technical users get a faster, scriptless way to reach real hardware and generate test code from the tools they already work in; non-technical users get direct access to device testing and test creation that previously required an automation engineer.

Developers test and debug on real devices without leaving their AI assistant, and generate runnable test code for the features they build—exporting authored test cases in the language and framework they already use.
QA engineers automate device selection, bug reproduction, and evidence capture, and author, run, and schedule test cases and suites from a prompt, so setup work disappears and they can focus on test logic.
Product managers verify shipped features and triage customer escalations on real hardware in minutes, and create tests by describing them in plain language—no engineer or code required.
Designers review layouts across real devices, locales, dark mode, and text sizes, and turn what they want to check into runnable test cases without writing code.
Support and QA leads confirm device compatibility and run quick smoke checks on demand, removing a common bottleneck.