Skip to main content
LoginSign Up
Products
AI Device CloudLiveOverviewReal iPhones & iPadsReal AndroidDesktop BrowsersEmulators & SimulatorsReal Device CloudApp LiveAndroid EmulatorsiOS SimulatorsDevice Coverage StrategymacOS Browser TestingWindows Browser TestingSafari Browser TestingLocal TestingGeolocationLanguage & TimezoneNetwork ThrottlingDevice AudioSession CaptureBrowser DevToolsNetwork LogsApp InspectorBug ReportingFiles & MediaAutomateOverviewSeleniumPlaywrightAppiumEspressoXCUITestParallel TestingPer-Test ArtifactsSession RecordingConsole & Network LogsMobile Network CaptureBuild HistoryCI/CDAI Device CloudTest ManagementTest Case ManagementJira PluginChrome Extensionnpm PackageTest CasesTest CyclesRequirementsDefectsReports & InsightsAI FeaturesProject ManagementJiraGitHubGitLabTestKase Built-inCI/CDWebhookVisualVisual Testingnpm PackageBaselinesIgnore RegionsReview WorkflowScheduled RunsAccessibilityAccessibility TestingChrome ExtensionWeb ScannerWorkflow AnalyzerAuthenticated ScanningTeam SharingFix PlansProgress Over TimeScheduled ScansWCAG 2.2 CoverageWCAG 2.2 ChecklistEuropean Accessibility ActADA ComplianceSection 508EN 301 549VPAT and ACR
Pricing
Developers
Product DocumentationTest Management APIDeviceFarm APICapability Generator
MCP servers

Two MCP servers for the assistant you already use

TestKase has two MCP servers. The Test Management MCP works with your test cases, cycles, plans and reports. The Device Farm MCP opens real phones and desktop browsers and uses them. Connect Claude, GitHub Copilot, Cursor or any MCP client with one access key.

Get started freeSee both servers

One key for both · the key stays on your computer · no code to write

Your AI assistantClaude · Copilot · Cursor
“Run the login test on a real phone, then record the result in my test cycle”
MCP

Test Management MCP

11 tools

  • search_testcases
  • manage_testcase
  • execute_tests
  • get_report
  • …and 7 more

Device Farm MCP

26 tools

  • start_device
  • read_screen
  • tap
  • analyze_failure
  • …and 22 more

Two servers. Pick one or use both.

One manages your tests. The other drives real devices and browsers. Each one installs on its own, and the same access key works for both.

TestKase Test Management MCP

@testkase/test-management-mcp11 tools

Your assistant works with your test cases, folders, cycles, plans and reports. It records results, but it does not run tests.

What it does

  • Create, update and search test cases
  • Organise folders
  • Build and search test cycles and test plans
  • Record results, by hand or from CI
  • Pull any of 40+ reports: execution, coverage, trends, team, defects, insights, risk and release

Try asking

  • “Find every blocked test case in the Payments folder.”
  • “Write test cases for the login flow: email, SSO and 2FA.”
  • “Mark TEST-5 and TEST-6 as passed in TCYCLE-10.”
  • “Show me the coverage report for this project.”

Install

Add this to your assistant's MCP config file:

{
  "mcpServers": {
    "testkase": {
      "command": "npx",
      "args": ["-y", "@testkase/test-management-mcp"],
      "env": {
        "TESTKASE_PAT_TOKEN": "xyz_your_token_here"
      }
    }
  }
}

Or, in Claude Code, run:

claude mcp add testkase -e TESTKASE_PAT_TOKEN=<key> -- npx -y @testkase/test-management-mcp
Read the docs View on npm

TestKase Device Farm MCP

@testkase/device-farm-mcp26 tools

Your assistant opens a real phone or a desktop browser in the TestKase AI Device Cloud and uses it. It can also read your Automate runs.

What it does

  • Open a real Android phone or iPhone, with your app or a web page
  • Open Chrome, Edge, Firefox or Safari on Windows or macOS
  • Read the screen as a numbered list, then tap, type, swipe, press keys and open URLs
  • Take screenshots and run several steps in one call
  • Read Automate runs: tests, commands, logs, video and screenshot links, and why a run failed
  • Give Selenium, Appium and Playwright connection settings
  • Upload app builds (.apk, .ipa, simulator .zip), check Local Tunnel status and stop a stuck run

One device at a time. A session ends after about 3 minutes idle or 60 minutes in total. It uses your Automate minutes.

Try asking

  • “Open our sign-up page in Safari on macOS and take a screenshot.”
  • “Install my latest .apk on an Android phone and log in as the test user.”
  • “Why did my last Automate run fail?”
  • “Give me the Playwright connection settings.”

Install

Add this to your assistant's MCP config file:

{
  "mcpServers": {
    "testkase-device-farm": {
      "command": "npx",
      "args": ["-y", "@testkase/device-farm-mcp"],
      "env": {
        "TESTKASE_PAT_TOKEN": "xyz_your_token_here"
      }
    }
  }
}

Or, in Claude Code, run:

claude mcp add testkase-device-farm -e TESTKASE_PAT_TOKEN=<key> -- npx -y @testkase/device-farm-mcp
Read the docs View on npmAbout AI Device Cloud

Better together

Add both servers and your assistant can test on a real device and then write down what happened. You ask once. It uses the right server for each part.

“Run the login test on a real phone, then record the result in my test cycle.”
  1. 1

    Device Farm MCP runs it

    Opens a real phone, goes through the login steps and checks the screen.

  2. 2

    Test Management MCP records it

    Marks the login test as passed or failed in your test cycle.

Set up in three steps

  1. Step 1

    Get your access key

    In TestKase, open Profile, then API Keys, and create a key. It starts with xyz_. One key works for both servers.

  2. Step 2

    Add the config

    Paste the config block for the server you want into your assistant's MCP config, with your key in place of the placeholder. In Claude Code, run the one-line command instead.

  3. Step 3

    Restart and ask

    Restart your assistant so it loads the server. Then ask it to do something in plain words.

Full setup guides:Test Management MCP Device Farm MCP

Any MCP client

Works with Claude Desktop, Claude Code, GitHub Copilot in VS Code, Cursor and any other MCP client.

One key for both

The same TestKase access key connects both servers. You do not need a second key.

Your key stays with you

The key lives in the config on your own computer. You never paste it into a chat.

FAQ

Frequently asked questions

MCP, the Model Context Protocol, is an open standard that lets an AI assistant use an outside tool directly. An MCP server is the small connector that gives the assistant those tools. TestKase has two: one for test management and one for the device farm.

Read the guide: Playwright MCP vs CLI vs test agents →

Connect your assistant in a few minutes

Create an access key, paste a config block, restart your assistant and ask.

Get started freeSee both servers