AutoGenesis MCP
Microsoft MCP servers for test automation: control of Windows, macOS and mobile UIs and generation of BDD code.
- 138 ★ on the repository
- Last commit
Broad access
Why this level
- The agent drives the mouse and keyboard in applications and can press buttons and change data in the program under test.
- The server sends usage data to Microsoft telemetry by default.
- Mobile tests through BrowserStack go out to an external cloud.
Install
- Works in
Manual install
git clone https://github.com/microsoft/AutoGenesis.gitThere is no ready package. Then run `uv sync` in the chosen server folder and start `uv run python simple_server.py --transport sse`.
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- The agent drives the mouse and keyboard in applications and can press buttons and change data in the program under test.
- The server sends usage data to Microsoft telemetry by default.
- Mobile tests through BrowserStack go out to an external cloud.
This is third-party code. Review the repository files before installing.
What it does
AutoGenesis lets an agent drive an application's interface and record the steps as tests. The repository has two servers: `pywinauto-mcp-server` automates Windows programs through the pywinauto library, and `appium-mcp-server` serves macOS and iOS and Android apps, including the BrowserStack cloud. The agent performs scenario steps through tool calls and the server saves the step code for Behave. The `.github/skills` folder holds the `autoGenesis-run` and `autoGenesis-win` skills: the first reads a scenario from a `.feature` file and runs it step by step with retries, the second handles Windows. A BDD AI Toolkit extension for VS Code covers recording and replay.
Who it is for. QA engineers and developers who need automated tests for desktop and mobile apps from natural-language scenarios.
Good fit when
- A Windows desktop program with no API needs an automated test built from described steps.
- A scenario sits in a `.feature` file and has to run on a mobile app.
- A macOS app is checked through Appium with the Mac2 driver.
- A tester records actions in VS Code and wants one-click replay.
Not a fit when
- A web page is under test, where Playwright fits.
- The project license must be unambiguous: GitHub did not detect its type.
- Usage data must not go to external telemetry.
Example request
Run the "Log in to the app" scenario from `demo.feature` and save the generated steps into `behave-demo/features/steps/`.MCP
- Transport
- stdio, sse
- Authentication
- not required
Limitations
The servers run from a clone of the repository through `uv`, with no ready-made package to install. The Windows server runs only on Windows with Python 3.10 or newer. macOS needs Xcode, Appium and the Mac2 driver, and mobile tests need BrowserStack. By default AutoGenesis sends usage data to the project's Application Insights instance. The app and its executable path are defined in `conf/pywinauto_conf.json`. The last commit dates to October 2026.
How to disable. Remove the MCP entry from `.vscode/mcp.json` or Cursor settings and stop the server process. Delete the repository clone by hand.
FAQ
How do I start the AutoGenesis Windows server?
Clone the repository, run `uv sync` in `pywinauto-mcp-server`, then `uv run python simple_server.py --transport sse`. It uses SSE on port 8000 by default.
How do I connect the server to VS Code?
Add `http://localhost:8000/sse` to `.vscode/mcp.json`, or use stdio mode through `uv run`, which VS Code starts itself. For Cursor, set it up in its MCP settings.
Which platforms are supported?
Windows through pywinauto, macOS through Appium with the Mac2 driver, and iOS and Android through Appium with the BrowserStack cloud.
What does the autoGenesis-run skill do?
It reads a scenario from a `.feature` file in `behave-demo/features/`, performs the steps with MCP calls and retries, and saves the code through `preview_code_changes` and `confirm_code_changes`.
Can telemetry be disabled?
The documentation does not describe it. Usage data goes to the project's Application Insights by default, so ask the authors how to turn it off.
Discussion
No questions or reviews yet.
Related
MCP serversWebdriverIO MCP: browser and mobile app automation
webdriverio-mcp
The official MCP from the WebdriverIO team drives Chrome, Firefox, Edge, Safari, Electron, iOS and Android through one tool set
MCP server, agent and skills for end-to-end UI tests where the agent drives the computer through vision on cloud machines
CLI, MCP server and typed Node.js API so an agent can verify its changes in a live mobile app on iOS, Android and more
BrowserStack MCP Server: testing on real devices
BrowserStack MCP Server
The official BrowserStack MCP server: real devices, automated tests and test cases driven by an agent from your editor
The agent finds Robot Framework keywords, runs steps live on a site, API or app and saves a ready .robot suite.
kogiQA MCP gives an agent a browser that finds page elements without selectors, for debugging web apps and writing e2e tests
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| AutoGenesis (this card) | yes | not required | 138 | |
| Windows-MCP | no | not required | 11,574 | general Windows control for an agent; AutoGenesis targets tests and saves steps as Behave code |
| Mobile MCP | no | not required | 10,529 | drives mobile devices; AutoGenesis adds Windows, macOS and BDD scenario generation |
README in short
Routes readers by platform: Windows, macOS, mobile apps and the VS Code extension. Gives the folder layout, telemetry details and a feedback contact.