MCP Server

The MCP Server gives AI assistants direct access to performance auditing, analysis, and reporting — right from your editor or chat interface.

Built on the Model Context Protocol, the server exposes a set of tools that any MCP-compatible client can call. This means you can run Lighthouse audits, analyze Core Web Vitals, and generate PDF reports without leaving the conversation.

What is MCP?

The Model Context Protocol (MCP) is an open standard that lets AI assistants connect to external tools and data sources. Instead of copy-pasting results between tabs, MCP lets your AI client call tools directly — passing parameters, receiving structured data, and acting on the results in context.

The MCP Server implements this protocol, turning your AI assistant into a performance engineering co-pilot.

Tools

The server exposes five tools:

Get Audit

Retrieves an existing Lighthouse audit by ID. Use this to pull up previous results, compare against a baseline, or feed historical data into a follow-up analysis.

Run Audit

Triggers a new Lighthouse audit against a given URL. Returns the full audit payload including performance scores, diagnostics, and resource timing data. This is your starting point for any performance investigation.

Run AI Analysis

Runs an AI-powered analysis on your audit data. You can target a specific Lighthouse category (Performance, Accessibility, Best Practices, SEO) or drill into a single Core Web Vital metric (LCP, CLS, INP, FCP, TTFB). The analysis surfaces actionable insights and prioritized recommendations based on your specific results.

Get CrUX Data

Fetches real-user Chrome UX Report (CrUX) data for a given URL or origin. This gives you field data — actual user experience metrics collected from real Chrome sessions — to complement the lab data from Lighthouse audits.

Generate Audit Report PDF

Generates a formatted PDF report from an audit. Useful for sharing results with stakeholders, attaching to tickets, or archiving for compliance purposes.

How It Works

1. Connect the MCP server to your preferred client (see setup guides below)
2. Ask your AI assistant to run an audit, analyze results, or generate a report
3. The assistant calls the appropriate tool, receives structured data, and responds in context

There's no dashboard to navigate, no UI to learn. You interact with the performance tools through natural language in the client you're already using.

Setup Guides

Choose your client to get started:

Each guide includes a video walkthrough and step-by-step instructions.