TestDriver MCP
MCP server, agent and skills for end-to-end UI tests where the agent drives the computer through vision on cloud machines
- 356 ★ on the repository
- Last commit
Broad access
Why this level
- Performs real UI actions and can run commands
- Sends the app and data to the vendor cloud
- Requires a secret access key
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add mcp/testdriverDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewBroad accessThe tool can change data, run commands or handle secrets.
- Performs real UI actions and can run commands
- Sends the app and data to the vendor cloud
- Requires a secret access key
This is third-party code. Review the repository files before installing.
What it does
The repository ships an MCP server, an expert agent and a set of skills for Claude Code and Codex to author and run end-to-end tests by driving the interface. Tests are stated in plain language: find an element, click, assert that something is visible, and the engine relies on vision rather than selectors. A run happens on an ephemeral cloud machine that launches a browser or app and returns a result with a video replay. A CLI sets up the project and wires it into Vitest, and there is a standalone npm package.
Who it is for. For QA engineers and developers who need end-to-end tests of hard flows: desktop apps, extensions, OAuth, PDF, iframe and canvas.
Good fit when
- You need to cover a flow that is hard to test with selector-based tools
- You want to describe test steps in words and run them from the agent
- You need a cloud run with a video replay and a report
Not a fit when
- Plain unit or selector tests without a cloud are enough
- You cannot expose the app and data to an external service
Example request
Write a test: open the site, sign in via OAuth and check that the username is visibleMCP
- Transport
- stdio
- Authentication
- API key
Environment variables
TD_API_KEYrequired, secret- Access key for the TestDriver cloud from the account console.
Limitations
It needs a TestDriver account and a TD_API_KEY, runs happen on the vendor's cloud machines, and billing follows their terms. Availability from Russia without a VPN is not verified. Tests perform real UI actions and can run commands, so they belong in test environments rather than production.
How to disable. Uninstall the testdriver plugin via /plugin, remove the testdriver MCP block from the agent config, and drop the testdriverai package from the project.
FAQ
Do I need a paid account?
You need a TestDriver account and a TD_API_KEY, and runs happen in their cloud. Sign-up is free; further limits and pricing are on the vendor's side.
How is it different from Playwright or Cypress?
A test is written in words and relies on computer vision rather than selectors, so it covers desktop apps, extensions, OAuth, PDF, iframe and canvas.
Discussion
No questions or reviews yet.
Related
MCP serversAn MCP server that spins up several browser agents to test a site for broken links, UI bugs and accessibility issues
TesterArmy CLI and MCP run end-to-end browser and mobile tests in the cloud from plain-language steps.
QA agents, skills and an MCP server: test generation, coverage gaps, flaky tests, security scans and chaos testing
kogiQA MCP gives an agent a browser that finds page elements without selectors, for debugging web apps and writing e2e tests
An agent gets a live session of your web, iOS or Android app, sees the screen, writes tests and checks its own edits.
An MCP server that drives the 1C test client's interface directly: opens forms, fills fields, records and replays scenarios
README in short
The README presents TestDriver as an end-to-end testing tool where the agent drives the interface through vision. Quick start: npx testdriverai init creates the project, installs Vitest and configures MCP for VS Code, Cursor or Claude Desktop, then vitest run executes a test on a cloud machine with a video replay. The repository holds an MCP server, an agent, dozens of per-action skills, a CLI and an editor extension. It claims support for desktop apps, browser extensions, OAuth, PDF, iframe and canvas.