How are tests defined?
Tests are JSON files containing an array of test objects, each with a name and an array of actions like `goto`, `click`, `type`, and `assert_text`. No JavaScript or imports required.
@fastslack
基本情報
AI-native E2E test runner. JSON tests, parallel Chrome pool, MCP integration with Claude Code. Paste a GitHub issue, get a runnable test back.
以下の設定を使って、このサーバーを MCP 対応クライアントに追加してください。
{
"mcpServers": {
"mtw-e2e-runner": {
"command": "npx",
"args": [
"skills",
"add",
"fastslack/mtw-e2e-runner"
]
}
}
}ツール一覧はまだありません
この掲載には表示できるツール一覧がまだありません。サーバーにツールがないという意味ではありません。対応機能はプロジェクトのドキュメントをご確認ください。
上のボタンで一時的なサンドボックスでこのサーバーを実行し、ツール一覧を取得できます。
Matware E2e Runner is an AI-native, zero-code browser testing framework where tests are plain JSON files. It integrates deeply with AI via an MCP server, enabling Claude Code to create, run, and debug tests conversationally. It is designed for teams that want to write and maintain end-to-end tests without learning Playwright, Cypress, or any test framework.
Install via npm install --save-dev @matware/e2e-runner, then run npx e2e-runner init to create a sample test directory, npx e2e-runner pool start to launch a shared Chrome pool in Docker, and npx e2e-runner run --all to execute tests. Configure your app’s port in e2e.config.js. Optionally add Claude Code integration with claude plugin marketplace add fastslack/mtw-e2e-runner to get 13 MCP tools.
Tests are JSON files containing an array of test objects, each with a name and an array of actions like `goto`, `click`, `type`, and `assert_text`. No JavaScript or imports required.
Node.js >= 20 and Docker must be running. Your application must be accessible on a known port, and Chrome executes inside Docker using `host.docker.internal` to reach your app.
An MCP server provides 13 tools to Claude Code, allowing it to create tests, run them, capture screenshots, visually verify pages, and debug failures — all from a chat interface.
Tests run in parallel against a shared Chrome pool managed via Docker (browserless/chrome). Serial mode is available for tests that share state.
Yes. Test-level and action-level retries are configurable. CI output includes JUnit XML, exit code 1 on failure, and auto-captured error screenshots.

Real UI references for AI coding agents: 500,000+ real screens, 15k+ website design systems with a DESIGN.md, and 30k+ UI elements, so your agent builds UI that looks real, not AI slop.

Catch intent conflicts before code conflicts. The open-source coordination protocol for coding agents, built above Git.


Agent-Native Terminal

The tools your AI needs to make your games. Turn prompts into production-ready game assets — animated characters with their weapons attached, seamless textures, terrain, VFX, HUDs — and playable game kits Unity, Godot, U
コメント