Seven third-party script patterns we triage before touching CSS
Before we debate font weights and hero layouts, we check these script patterns that quietly move LCP and INP in production.
Series
Technical how-tos
Craft and trade-offs: CI, budgets, scripts, and implementation detail.
Posts in this series
Manual PageSpeed runs are honest; they also do not scale
Spot checks stay useful for one URL and one decision. They stop being a strategy when the client list grows.
Portfolio dashboards: what multi-tenant changes day to day
Multi-tenant is not a badge on a pricing page. It is how your team stops re-learning a new monitoring surface for every client and starts asking portfolio questions in standup
What performance checks belong in CI (and what belongs in monitoring)
Lighthouse CI can block merges or become a job teams ignore. Here is how we split pre-merge checks, bundle limits, and scheduled PageSpeed monitoring so CI only tests what it can do reliably.
Lab green, field amber: reconciling two PageSpeed truths
Lighthouse can look healthy while CrUX still needs improvement. Here is how we decide which signal to trust when the two disagree.
MCP versus API: what assistants need that your REST endpoints do not spell out
Your HTTP API returns JSON; Model Context Protocol exposes tools, resources, and prompts for ChatGPT and Claude. When to ship both, and when MCP is not a replacement for REST.
