Playwright Lighthouse MCP Server

Playwright‑Lighthouse MCP Server is a lightweight, protocol‑driven service that bridges the gap between large language models and real‑world web performance analysis. By exposing a set of tools over the Model Context Protocol (MCP), it lets AI assistants such as Claude execute automated Lighthouse audits and capture page screenshots without leaving the conversational flow. This eliminates the need for developers to manually run command‑line tools or write custom integrations, allowing performance insights to be generated on demand directly from an LLM interface.

The server’s core functionality centers around two high‑value tools: and . launches a Playwright instance, navigates to the supplied URL, and runs a Lighthouse audit for one or more categories (performance, accessibility, best‑practices, SEO, PWA). It returns a concise report that highlights overall scores, color‑coded indicators, and the most actionable improvement items per category. captures a visual snapshot of the page, optionally rendering the full scrollable area. These outputs are serialized in JSON and stored locally, with the report path included for easy reference.

For developers building AI‑augmented workflows, this server offers a plug‑and‑play model: simply add the MCP configuration to your client (e.g., Claude for Desktop), and the assistant can invoke or as if they were native commands. This opens up a range of use cases, from automated performance regression testing in CI pipelines to real‑time diagnostics during user interviews. Teams can ask the assistant, “How fast is our landing page?” and receive a Lighthouse score with suggested fixes, all without leaving the chat.

Unique advantages stem from the tight coupling of Playwright’s headless browser control with Lighthouse’s comprehensive audit engine. The server can capture the exact state of a page—including dynamic content loaded via JavaScript—ensuring that performance metrics reflect what real users experience. Additionally, the MCP interface keeps the tool’s internals hidden from end‑users while still allowing fine‑grained parameter control, such as selecting specific categories or limiting the number of improvement items. This balance of power and simplicity makes Playwright‑Lighthouse MCP a compelling addition to any AI‑driven development toolkit.