What it does

The @letsrunit/mcp-server bridges AI agents and letsrunit's browser testing framework via the Model Context Protocol. It enables agents to generate, write, run, and debug Gherkin/Cucumber tests against web applications in a real browser. Tests live as plain .feature files compatible with Cucumber and can run in CI pipelines after the agent session ends.

Who it's for

AI coding agents (Claude Code, etc.) building test suites for web applications, and developers who want to offload browser test automation to agents. Most useful when agents need to verify UI behavior, generate test coverage, or debug failing tests against a live browser.

Common use cases

Generate browser tests via agents using plain-language Gherkin syntax
Run Cucumber feature tests against Chromium during development or in CI
Debug failing tests by having agents inspect and fix Gherkin scenarios
Test email-based workflows using the bundled mailbox integration
Keep test suites alive in CI after agents finish a session

Setup pitfalls

Must run npx letsrunit init to configure MCP, Playwright/Chromium, and environment variables
Requires LETSRUNIT_BASE_URL environment variable pointing to the app under test
Needs filesystem write access to create and manage .feature files and Cucumber configurations
Browser automation adds significant disk and memory overhead; init automates setup but verify your environment has adequate capacity