Learn to diagnose and resolve web performance issues using Google Gemini and Chrome DevTools MCP.
- 1: Introduction and Anatomy of the MCP (Tools exposed to the LLM).
- 2: Gemini + Chrome DevTools MCP Configuration.
- 3: Exercise 1: Manual Analysis. Detecting performance problems without AI help.
- 4: Exercise 2: The Minified Code Wall. Using MCP in production environments (without Source Maps).
- 5: Exercise 3: From Trace to Code. Using Source Maps to identify the exact problem.
- 6: Exercise 4: Total Automation. Branch, Commit, and PR with the performance fix.
- 7: Exercise 5: Deep Analysis and External SKILLs. Expert knowledge and best practices from Vercel/Next.js.
- 8: Q&A and Workflows applicable to real projects.
docs/: Installation guides, technical concepts, glossary (reference.md), and advanced workflows (workflows.md).exercises/: Step-by-step guides for each of the workshop phases.01-manual-analysis/: Traditional analysis without AI help.02-mcp-minified/: Using MCP in production environments (obfuscated code).03-sourcemaps/: Mapping to the original source code.04-git-workflow/: Automating fixes with branches and commits.05-deep-analysis/: Expert analysis with Vercel SKILLs in Next.js.
demos/: Example projects with intentional bottlenecks.performance-lab/: Basic HTML/JS project (LCP and Long Tasks).sourcemaps-app/: Vite/React app for debugging with Source Maps.nextjs-performance-app/: Next.js app for static analysis and best practices.
- Google Gemini (or an MCP-compatible client).
- Node.js installed to run the MCP server and demos.
- Basic knowledge of Chrome DevTools and Core Web Vitals.
Check the detailed guide at docs/setup.md to get started.
![Build [>] debug & deploy with AI](/nucliweb/webperf-debugging-devtools-mcp/raw/main/assets/geminicli.png)