Lighthouse MCP Server

The Lighthouse MCP Server turns the powerful, industry‑standard web audit tool Google Lighthouse into a first‑class AI‑ready service. By exposing Lighthouse’s rich audit data through the Model Context Protocol, it lets large language models and autonomous agents query real‑world websites for performance, accessibility, SEO, security, and resource usage insights—all in a single API call. This removes the need for developers to manually run Lighthouse, parse JSON reports, or embed complex browser tooling into their own pipelines.

At its core, the server accepts a URL and optional device or throttling parameters, then orchestrates a Lighthouse run inside a managed Chrome/Chromium instance. The resulting report is transformed into an MCP‑friendly format: structured metrics, pass/fail flags against thresholds, and actionable recommendations. Developers can then ask an AI assistant to “compare the performance of on mobile versus desktop” or “summarize accessibility issues in the latest audit,” and receive concise, context‑aware answers that include raw scores, visual highlights, and links to best‑practice resources.

Key capabilities include:

• Comprehensive audits covering Core Web Vitals, accessibility (WCAG), SEO best practices, and security checks such as HTTPS status and CSP validation.
• Device‑aware analysis with built‑in mobile/desktop comparison and network throttling options, enabling realistic performance budgeting.
• Custom budgets that let teams enforce thresholds for metrics like LCP or total bytes, turning audit results into policy checks.
• Rich resource profiling that identifies heavy JavaScript bundles, oversized images, or redundant fonts, helping teams target optimization work.
• Integrated reference material that links recommendations to official guidelines, making the output immediately actionable.

Real‑world use cases span from continuous integration pipelines that automatically flag regressions in Core Web Vitals, to chat‑based development assistants that can explain audit findings without the user leaving their IDE. For example, a developer could type “What are the biggest performance bottlenecks on ?” into VS Code, and the AI would return a Lighthouse‑derived summary complete with actionable steps. Similarly, content creators can ask an assistant to audit a landing page for SEO compliance and receive a prioritized list of improvements.

Because the server is built on Node 22+ and bundles Chrome internally, it requires minimal setup beyond installing the MCP package. It integrates seamlessly with any MCP‑compliant client—VS Code, Claude Desktop, Cursor, or custom agents—making it a plug‑and‑play component in modern AI‑augmented development workflows. The Lighthouse MCP Server therefore empowers developers to surface deep, trustworthy web audit data directly into the tools they already use, accelerating quality assurance and reducing the friction of manual performance engineering.