Zero-config, fully remote MCP server that gives AI agents end-to-end browser testing, screenshots and CLI tooling.
https://github.com/debugg-ai/debugg-ai-mcpYour AI agent can now run comprehensive browser tests without touching a single line of Playwright code.
E2E testing shouldn't be this hard. You spend more time wrestling with browser automation setup than actually testing your features. Chrome popups interrupt your flow. Playwright versions break between teammates. CI environments need constant browser binary updates.
Meanwhile, your localhost app changes daily, but your test suite doesn't keep up because writing and maintaining E2E tests is frankly a pain.
This MCP server flips the script entirely. Instead of writing test scripts, you tell your AI agent what to test in plain English:
The server handles everything else - browser provisioning, test execution, screenshot capture, and detailed reporting. All running remotely so your local development stays clean.
Zero Configuration: No Playwright setup, no browser management, no proxy configuration. Just grab an API key and add one entry to your MCP server config.
Fully Remote Execution: Tests run on managed remote browsers. No more Chrome popups disrupting your flow or version conflicts between team members.
Natural Language Interface: Your AI agents can understand test requirements and execute them without you writing a single assertion or selector.
Instant Feedback: Get real-time progress updates and final screenshots showing exactly what happened during the test run.
Historical Test Tracking: All test results automatically appear in your Debugg AI dashboard for later analysis or CI/CD integration.
During Development: Push a new feature branch, ask Claude to "test the new signup flow" and get immediate feedback on whether it works end-to-end.
Code Review Automation: Have your AI agent automatically test PR changes by describing the expected functionality in natural language.
Regression Testing: Quickly verify that existing flows still work after code changes without maintaining brittle test scripts.
QA Handoff: Generate comprehensive test documentation by having the AI describe what it tested and what it found.
Add this to your Claude Desktop config:
{
"mcpServers": {
"debugg-ai-mcp": {
"command": "npx",
"args": ["-y", "@debugg-ai/debugg-ai-mcp"],
"env": {
"DEBUGGAI_API_KEY": "your_api_key",
"DEBUGGAI_LOCAL_PORT": 3000
}
}
}
}
Then simply ask Claude: "Test my login page at localhost:3000" and watch it work.
This isn't about replacing your existing test infrastructure - it's about making testing accessible during development when you need quick feedback. Run it locally while coding, use the dashboard for organized test suites in CI/CD.
The server connects to your localhost app via secure tunnels, so you're testing your actual running code, not some deployed version that might be stale.
Get started: Grab a free API key at debugg.ai and add the MCP server to your setup. Your next feature can have comprehensive E2E tests running in under 5 minutes.