What does this skill do?

Skill Performance Benchmark provides a systematic method for measuring, comparing, and monitoring the performance of web applications, APIs, and build processes. It establishes reliable baselines, runs measurements before and after each change, and issues an automatic verdict for each metric, so that regressions are detected before they reach production.

Before-and-After Comparison of PR
Run a baseline measurement, apply the PR changes, and automatically compare metrics such as LCP, bundle size, or build time, with an immediate result.
Website Performance
Measures Core Web Vitals (LCP, CLS, INP, FCP, TTFB), resource size, and network requests using the MCP browser for each target URL.
API Latency and Load
Send 100 requests to an endpoint, measure the p50, p95, and p99 values, test for concurrency, and verify SLA compliance.
Build Times and CI
Measure cold builds, hot reloads, test suites, TypeScript validation, linting, and Docker builds to identify bottlenecks.

Usage examples

🔀 Compare a PR
I just applied the changes from my PR. Run a benchmark comparison against the baseline I saved earlier and let me know if there are any regressions.
🌐 Measure Core Web Vitals
Measures the performance of https://miapp.com in page mode: LCP, CLS, INP, FCP, TTFB, and resource size.
🔌 API Latency Test
Send 100 requests to the /users endpoint with 10 concurrent connections and check whether the p95 meets the 500-millisecond SLA.
🏗️ Measure build times
Measure the cold build time, Hot Reload, tests, TypeScript, and linting for my project to identify bottlenecks.

Features

Automatic verdict based on metrics Each metric receives a verdict (BETTER, WARNING, or REGRESSION) based on the absolute delta relative to the baseline.
Versionable baselines in Git Baselines are saved as JSON files in .ecc/benchmarks/ and can be versioned so that the entire team can share references.
Complete Core Web Vitals It measures LCP, CLS, INP, FCP, and TTFB with clear objectives, as well as bundle weight, images, and third-party scripts.
API Analysis Under Load Run 100 requests with configurable concurrency and get the P50, P95, P99, response size, and status codes.
CI Integration Add a benchmark comparison to your pipeline to run it on every PR and keep performance under constant control.

Frequently asked questions

For web page performance mode, you need an MCP browser that allows you to navigate to URLs and measure actual browser metrics. API and build modes work with system commands without an additional MCP.
Baselines are stored as JSON files in the .ecc/benchmarks/ folder within your project. You can version them in Git so that the entire team shares the same references.
Yes. Run a baseline benchmark to establish a baseline and a comparison benchmark for each PR. The automatic verdict allows you to block merges if there are regressions.
Mide LCP (<2.5s), CLS (<0.1), INP (<200ms), FCP (<1.8s) y TTFB (<800ms), además del peso total de página, bundle JS, CSS, imágenes, scripts de terceros y peticiones de red.
Performance Benchmark — Detecting Regressions with Claude AI

¿Prefieres escuchar el contenido? Genera la narración de audio con un clic.