Alumnium MCP
Alumnium gives an agent end-to-end tests over MCP, with Playwright, Selenium, Appium or Maestro running the app steps.
- 1,022 ★ on the repository
- Last commit
Extended access
Why this level
- Page content is sent to the model provider.
- The installer runs through curl and sh.
Install
- Works in
Manual install
curl -LsSf https://alumnium.ai/install.sh | sh
claude mcp add alumnium --env OPENAI_API_KEY=... -- alumnium mcpThe first command of the Installation section installs the CLI, the second registers MCP in Claude Code.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Page content is sent to the model provider.
- The installer runs through curl and sh.
This is third-party code. Review the repository files before installing.
What it does
Alumnium is an MCP server and library for end-to-end testing. The agent gets an application address, performs steps in a browser or on a mobile device and verifies the outcome. Familiar drivers do the work underneath: Appium, Maestro, Playwright and Selenium. The same actions exist in Python, TypeScript and Java code: `al.do(...)` performs a natural-language step, `al.check(...)` verifies a claim, `al.get(...)` reads a value from the page.
Who it is for. QA engineers and developers who ask Claude Code to check a web or mobile app and want assertions rather than only screenshots.
Good fit when
- After a UI change an agent walks the main scenario and reports what broke.
- Selenium or Playwright tests are extended with natural-language steps.
- A mobile scenario runs through Appium or Maestro without hand-written locators.
- You need assertions such as page title contains a word in Python, TypeScript or Java.
Not a fit when
- Tests must run without calling an external model.
- You need deterministic checks with fixed locators.
- Pages hold data that must not be sent to a model provider.
Example request
Open http://localhost:3000, go through sign-up and check that the user name appears after login.MCP
- Transport
- stdio, http
- Authentication
- API key
Environment variables
OPENAI_API_KEYrequired, secret- Model provider key used by Alumnium (README example)
Limitations
A model provider key is required, the README examples use `OPENAI_API_KEY`. Page content goes to the model, so avoid running tests on real personal data. The installer runs through `curl ... | sh`, so read the script first. Natural-language steps are less reproducible than classic tests. Native mobile checks need driver environments: an emulator or Appium server, or a device for Maestro.
How to disable. Remove the server with `claude mcp remove alumnium` and delete the `alumnium` binary. Take the provider key out of the environment.
FAQ
How do I connect Alumnium to Claude Code?
After installing, run `claude mcp add alumnium --env OPENAI_API_KEY=... -- alumnium mcp`. For other clients see the MCP documentation on the project site.
Which drivers are supported?
Appium, Maestro, Playwright and Selenium.
Is there an HTTP mode?
Yes. The README points to the MCP documentation for modes and running over HTTP.
Can it be used without MCP?
Yes. The library installs as `pip install alumnium`, `npm install alumnium` or a Java dependency.
Which languages do tests use?
Python (`pip install alumnium`), TypeScript (`npm install alumnium`) and Java through the `ai.alumnium:alumnium` dependency.
Discussion
No questions or reviews yet.
Related
MCP servershyperfine repeatedly runs commands, compares timing and other metrics and exports measurements for review and automation.
A CLI for testing and red teaming LLM applications: compares models, runs automated CI checks and finds vulnerabilities
The CLI runs Node.js tests concurrently and shows expected-versus-actual differences for an agent to investigate.
The CLI runs HTTP request chains from plain-text files and checks response status, headers and body content.
Open-source AI agent for pull request review: descriptions, feedback and improvement suggestions on GitHub, GitLab and more
Playwright CLI
playwright-cli
Microsoft's official Playwright CLI with an agent skill: drive a browser through short commands without heavy MCP schemas
Comparison with similar tools
Cards that share the main search query, and how they differ.
| Tool | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| Alumnium (this card) | no | API key | 1,022 | |
| Appium MCP | yes | not required | 518 | drives a device through Appium; Alumnium adds assertions and also works with Playwright and Selenium |
| MCP Server Playwright by Automata Labs | no | not required | 297 | exposes Playwright browser tools; Alumnium verifies natural-language assertions |
README in short
Setups are described for three languages and MCP. For checks the agent calls the same `do`, `check` and `get` as test code.