Cdn strategy

Skill Amey-Thakur/AI-SKILLS/skills/performance/cdn-strategy

Plug-and-play skills and prompts for every AI coding agent

Install

npx -y skills add Amey-Thakur/AI-SKILLS --skill cdn-strategy

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

18 days oldThe repository was created 18 days ago. New is not bad, but a brand new repository carrying a familiar-sounding name is the shape a typosquat arrives in, and there has been no time for anyone else to find a problem with it.
4 stars4 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Push work to the edge so requests never touch your origin, using cache headers, invalidation, and origin shielding. Use when origin load is high, global latency is uneven, or a CDN sits in front of your app but barely caches.

SKILL.md

2.9 KB, as published. Nobody here has run it

CDN strategy

A CDN only helps for the responses it actually stores. Most origins ship headers that quietly force every request back to the source, so the edge becomes an expensive proxy that adds a hop and caches nothing. The work is making assets and safe responses cacheable, then keeping them fresh.

Method

Read your cache hit ratio before touching config. Pull the edge hit ratio from your provider (Fastly Fastly-Debug, Cloudflare cf-cache-status, CloudFront X-Cache). Below 80% on static assets means headers, not capacity, are the problem. Fix the misses you can name first.
Separate immutable from mutable at the URL. Fingerprint static assets (app.4f1a.js) and serve them Cache-Control: public, max-age=31536000, immutable. HTML and API responses get short or zero TTLs. One caching policy per content class, keyed off the path, not per route guesswork.
Split browser TTL from edge TTL. Use s-maxage for the CDN and max-age for the browser so you can hold a response at the edge for an hour while telling clients to revalidate in a minute. stale-while-revalidate lets the edge serve stale bytes while it refetches in the background.
Invalidate by tag, not by guessing. Attach surrogate keys (Fastly Surrogate-Key, Cloudflare Cache-Tag) to responses so publishing an article purges article:912 everywhere in one call. Path purges miss query-string and locale variants; tag purges do not.
Turn on origin shielding. Designate one shield POP so the other edge nodes fetch from it instead of stampeding your origin on a cold cache. This collapses a global miss from hundreds of origin requests to one.
Control the cache key deliberately. Strip tracking query params and irrelevant cookies from the key, and set Vary only on headers that truly change the body (Accept-Encoding, sometimes Accept-Language). A Vary: Cookie on a public page fragments the cache into uselessness.

Litmus tests

Does curl -I on a static asset show a year-long max-age and immutable?
After a publish, does the changed page update within one purge call, with no stale variants left behind?
Does a cold-cache spike hit the origin once per shield, not once per POP?
Is cf-cache-status (or equivalent) HIT on the paths you expect to cache?

Boundaries

This covers edge caching of HTTP responses. Header semantics themselves belong to the http-caching skill; personalized or auth-gated responses that cannot be shared across users are a compute-at-edge or origin concern, not a caching one. Defer purge-on-deploy wiring to the project's release pipeline.