agentleFS
laguagulaguagu/claude-code-nextjs-skills/skills/cache-components/SKILL.md

cache-components

Skill72 starsChanged 17 days agoMITView on GitHub

Expert guidance for Next.js Cache Components and Partial Prerendering (PPR). Use when implementing 'use cache' directive, configuring cache lifetimes with cacheLife(), tagging cached data with cacheTag(), invalidating caches with updateTag()/revalidateTag(), optimizing static vs dynamic content boundaries, instant navigation validation, 'use cache: private', pass-through/interleaving patterns, GET Route Handler caching, debugging cache issues, and reviewing Cache Component implementations.

6 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use fullstack-developer, frontend-code-review and 4 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • belokonmSkill

    fullstack-developer

    End-to-end feature expert specializing in frontend-backend integration, system architecture, and complete application development

    Also in 3 projects that use this

  • chendongqiSkill

    frontend-code-review

    Trigger when the user requests a review of frontend files (e.g., `.tsx`, `.ts`, `.js`). Support both pending-change reviews and focused file reviews while applying the checklist rules.

    Also in 4 projects that use this

  • GoogleSkill

    pr-creator

    Also in 3 projects that use this

  • Hugging FaceSkill

    update-docs

    Update documentation across the repo after API changes. Finds stale references in docs, examples, docstrings, and fixes them.

    Also in 3 projects that use this

  • GoogleSkill

    code-reviewer

    Also in 3 projects that use this

  • VercelSkill

    find-skills

    Helps users discover and install agent skills when they ask questions like "how do I do X", "find a skill for X", "is there a skill that can...", or express interest in extending capabilities. This skill should be used when the user is looking for functionality that might exist as an installable skill.

    Also in 4 projects that use this

The file

We split this skill into the rules it gives an agent

6 rules. 0 are common, 3 are distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 6 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Keep the project's caching mode unless migration is requested.

    line 10
  2. Very specific:

    Read the relevant guide in node_modules/next/dist/docs/ first (bundled since 16.2.

    line 17
  3. Distinctive:

    Keep uncached I/O and request-only content below appropriate Suspense boundaries.

    line 134
  4. Distinctive:

    Use tags when invalidation must reach the same entity across routes, path invalidation for route-specific output, or expiry where that …

    line 146
  5. Very specific:

    Run the production build, then exercise cold/direct navigation, client navigation, the relevant mutation and subsequent read under next …

    line 165
  6. Distinctive:

    Check tenant isolation and multi-instance invalidation when applicable.

    line 167

“Common” is measured: kept by at least 6 in 10 of the 6 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Cache Components

Keep the project's caching mode unless migration is requested. Confirm the installed Next.js version and cacheComponents/partialPrefetching configuration before applying these patterns. Use Cache Components on next 16.4.1+: 16.0.0–16.4.0 allow unauthenticated RCE through use cache functions returning typed arrays (GHSA-673q-mjm6-r5xr; no workaround, upgrade even without such functions). Flag an affected install before cache work.

Read the relevant guide in node_modules/next/dist/docs/ first (bundled since 16.2; in a monorepo, use the app's own next package). Web paths map to files with numbered folders, so find them by name: .../functions/cacheLife is 01-app/03-api-reference/04-functions/cacheLife.md. Under 01-app/, start with 02-guides/migrating-to-cache-components.md, 01-getting-started/08-caching.md and 02-guides/instant-navigation.md. Error pages under /docs/messages are not bundled; next dev/next build print the fix options inline. Without local docs, use the official caching docs. Exact exports and signatures are also in node_modules/next/cache.d.ts. For static-output and staged-deferral questions in 16.4+, find the installed ensureStatic route-segment reference and navigation/prefetch function references by name.

The goal is a useful prerendered shell with freshness and authorization rules that remain correct after mutations, deploys and client navigation.

Facts older examples get wrong

  • Next.js 16 enables the mode with top-level cacheComponents: true. It replaces experimental.dynamicIO, experimental.useCache and experimental.ppr; cacheLife/cacheTag import from next/cache without unstable_. In 16.4, set partialPrefetching explicitly: leaving it unset warns; recommended create-next-app defaults enable both flags.
  • With it enabled, segment exports dynamic, revalidate, fetchCache, dynamicParams and runtime (even 'nodejs') are errors; only the default Node.js runtime is supported. revalidate = N becomes cacheLife (nearest or custom profile) inside use cache; force-dynamic and fetchCache are unnecessary; fetch next: { revalidate, tags } becomes cacheLife/cacheTag in a use cache function; request data goes under Suspense; dynamicParams = false becomes notFound() for unknown params.
  • Existing fetch and unstable_cache caching still works as a separate layer, so do not mechanically rewrite every read. Its persistence across deployments and instances depends on retained/shared storage; self-hosted instances do not share it by default. use cache is in-memory by default; even a durable handler cannot reuse entries when the build/deployment ID changes.
  • generateStaticParams must return at least one param: [] fails the build, and removing the export renders the route on every request.
  • revalidate: 0 or expire under 5 minutes makes a request-time hole instead of prerendered output; of the presets only seconds does. 16.3 adds stale thresholds (under 30 s: not prerendered; under 5 min: not in the App Shell).
  • Synchronous new Date(), Math.random() or crypto.randomUUID() during prerender is an error: capture it in use cache, or defer with await io() (next/cache, 16.3+) or await connection() (next/server) under Suspense.
  • use cache: private was experimental through 16.2. It runs at request time, stays out of the static shell and accepts no custom handler.
  • export const instant (16.3; unstable_instant in 16.2) opts a segment into instant-navigation validation. Since 16.3, every Page/Default segment is implicitly validated in next dev only (dev overlay) at the default level; build validation requires experimental.instantInsights.validationLevel set to experimental-error or experimental-manual-error (16.2 unstable_instant also validated at build). Test navigations with @next/playwright instant(). instant = false permits a blocking route; it does not clear synchronous-IO errors.

For whole-app adoption of Cache Components or Partial Prefetching, use the official workflow skills listed in the installed 01-app/02-guides/ai-agents.md (from vercel/next.js skills/); the bundled 02-guides/migrating-to-cache-components.md and adopting-partial-prefetching.md cover the same steps.

Shape of one route

app/posts/page.tsx, checked against next 16.3.8 with cacheComponents: true (tsc --noEmit, next build, the action under next start). @/lib/* is project code.

import { Suspense } from 'react'
import { cookies } from 'next/headers'
import { cacheLife, cacheTag, updateTag } from 'next/cache'
import { insertPost, listPosts } from '@/lib/posts' // project code
import { requireEditor } from '@/lib/auth' // throws unless the session may edit

async function getPosts() {
  'use cache' // shared entry: no cookies/headers/searchParams inside
  cacheLife('hours')
  cacheTag('posts')
  return listPosts()
}

async function Theme() {
  const theme = (await cookies()).get('theme')?.value // request-time, outside the cache
  return <p>Theme: {theme ?? 'system'}</p>
}

async function addPost(formData: FormData) {
  'use server'
  await requireEditor()
  const title = String(formData.get('title') ?? '').trim()
  if (!title) return
  await insertPost(title)
  updateTag('posts') // Server Actions only; the next read waits for fresh data
}

export default async function Page() {
  const posts = await getPosts() // cached, so it can be part of the static shell
  return (
    <>
      <ul>{posts.map((p) => <li key={p.id}>{p.title}</li>)}</ul>
      <Suspense fallback={<p>Theme: …</p>}><Theme /></Suspense>
      <form action={addPost}><input name="title" /><button>Add</button></form>
    </>
  )
}

Decide the boundary

  • Cache reusable results when the product's freshness and data policy permit. Static content needs no directive just to be static.
  • A plain use cache scope cannot read cookies, headers, searchParams or call connection(), including indirectly through helpers. Read those outside and pass authorized, serializable values. Account/tenant IDs must participate in the cache key when output depends on them. On a dynamically rendered route, the next-request-in-use-cache error can pass next build and surface only under next start.
  • Keep uncached I/O and request-only content below appropriate Suspense boundaries. Suspense supplies fallback UI; it does not itself make synchronous work dynamic.
  • Evaluate private/remote variants against installed docs and deployment handlers. Private caching is not a compliance guarantee; remote caching is not a consistency protocol.

Define freshness and invalidation

Choose lifetimes from actual acceptable staleness. Named profiles can be overridden by the project; read its config before assuming a duration.

Use tags when invalidation must reach the same entity across routes, path invalidation for route-specific output, or expiry where that meets the contract. Authenticate, authorize and validate before mutations. Invalidate only affected cached data:

  • updateTag: immediate expiry/read-your-writes, Server Actions only.
  • revalidateTag(tag, 'max'): stale-while-revalidate on a later visit.
  • revalidateTag(tag, { expire: 0 }): immediate expiry where a webhook/Route Handler needs it. The one-argument form is deprecated.

These tag APIs can also apply to tagged fetch data; their availability does not by itself mean Cache Components is enabled.

Read for the problem

  • API boundaries: keys, serialization, lifetimes, handlers and route APIs.
  • Composition: tenant isolation, nested caches, pass-through and dynamic params.
  • Troubleshooting: diagnose blocking, stale or inconsistent output.

Run the production build, then exercise cold/direct navigation, client navigation, the relevant mutation and subsequent read under next start. Check tenant isolation and multi-instance invalidation when applicable. A passing build cannot establish those behaviors.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.