Plugin Bundle Size logo

Plugin Bundle Size

Organization
grafana
plugin-bundle-size

Optimise Grafana app plugin bundle size using React.lazy, Suspense, and webpack code splitting. Use when the user asks to reduce plugin bundle size, optimise module.js, add code splitting, improve initial plugin load performance, split plugin chunks, lazy load plugin pages, or help implement lazy loading in a Grafana app plugin. Triggers on phrases like "optimise plugin bundle size", "module.js is too large", "plugin is slow to load", "code split the plugin", "reduce initial JS payload", or "help me with Suspense in my plugin".

Overview

Publishergrafana
Repositoryskills
Skill nameplugin-bundle-size
Stars
289
Forks
27
Bundled files
2
LicenseApache-2.0
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 2 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by grafana on GitHub. Read the source before you install it.

Installation

Install the Plugin Bundle Size AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/grafana/skills/tree/main/skills/grafana-plugins/plugin-bundle-size
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/grafana/skills.git /tmp/skills
mkdir -p .claude/skills
cp -r /tmp/skills/skills/grafana-plugins/plugin-bundle-size .claude/skills/plugin-bundle-size
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Plugin Bundle Size in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Plugin Bundle Size on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Plugin Bundle Size is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Grafana plugin bundle size optimisation

module.js is the render-blocking entry point for every Grafana app plugin. The smaller it is, the less impact the plugin has on Grafana's overall startup time. A well-split plugin should have a module.js under ~200 KB that contains nothing but lazy-loaded wrappers — all feature code loads on demand.

Target: ~15–25 JS chunks total. Fewer means too little splitting; far more (50+) means over-engineering.

Risk levels

Not all splitting opportunities carry the same risk. Apply them in this order:

LevelWhatRiskImpact
Safemodule.tsx lazy wrappers (Priority 1)Very low — no behaviour changeHighest — module.js drops 90%+
SafeRoute-level lazy() (Priority 2)Low — each route is self-containedHigh — one chunk per route
SafeExtension lazy() (Priority 3)Low — extensions are isolatedMedium — independent chunk per extension
ModerateComponent registries / tab panels (Priority 4)Medium — verify Suspense placementMedium — splits heavy pages further
Do not touchVendor libraries (@grafana/scenes, @reduxjs/toolkit)N/AN/A — webpack splits these automatically
Do not touchShared utility components (Markdown, Spinner) used across many filesHigh churn, many callsitesLow — already in shared vendor chunks

When in doubt, stop after Priority 2. Routes alone typically reduce module.js by 95%+.


Step 1: Add bundle size CI reporting (recommended)

Add the grafana/plugin-actions/bundle-size action to get automatic bundle size comparison comments on every PR. This posts a table showing entry point size changes, file count diffs, and total bundle impact.

Root-level plugins (plugin at repo root):

yaml
# .github/workflows/bundle-size.yml
name: Bundle Size
on:
  pull_request:
  push:
    branches: [main]
  workflow_dispatch:

jobs:
  bundle-size:
    runs-on: ubuntu-latest
    permissions:
      contents: write
      id-token: write
      pull-requests: write
      actions: read
    steps:
      - uses: actions/checkout@v4
      - uses: actions/setup-node@v4
        with:
          node-version-file: .nvmrc
      - name: Install and build
        run: yarn install
      - name: Bundle Size
        uses: grafana/plugin-actions/bundle-size@a66a1c96cdbb176f9cccf10cf23593e250db7cce # bundle-size/v1.1.0

Subdirectory plugins (e.g. plugin/ in a monorepo):

The action's install step runs at the repo root and cannot find yarn.lock in a subdirectory. Work around this by installing deps yourself and symlinking to root:

yaml
jobs:
  bundle-size:
    runs-on: ubuntu-latest
    permissions:
      contents: write
      id-token: write
      pull-requests: write
      actions: read
    steps:
      - uses: actions/checkout@v4
      - uses: actions/setup-node@v4
        with:
          node-version-file: ./plugin/.nvmrc
      - name: Install dependencies
        working-directory: ./plugin
        run: yarn install
      - name: Symlink plugin to root for bundle-size action
        run: |
          ln -s plugin/yarn.lock yarn.lock
          ln -s plugin/package.json package.json
          ln -s plugin/.yarnrc.yml .yarnrc.yml
          ln -s plugin/node_modules node_modules
      - name: Bundle Size
        uses: grafana/plugin-actions/bundle-size@a66a1c96cdbb176f9cccf10cf23593e250db7cce # bundle-size/v1.1.0
        with:
          working-directory: ./plugin

How it works: On push to main, builds and uploads a baseline artifact. On PRs, compares against it and posts a diff comment. Use workflow_dispatch to generate the first baseline.

Reference: grafana-k8s-plugin workflow


Step 2: Detect plugin context

bash
# Confirm this is an app plugin (type: "app" — datasource/panel plugins have different needs)
jq -r '"\(.id) — \(.type)"' src/plugin.json

# Locate the entry point
ls src/module.ts src/module.tsx 2>/dev/null

# Measure the current PRODUCTION bundle size BEFORE making any changes
# Dev builds are unminified and much larger — always measure production
yarn build 2>/dev/null || npm run build
echo "=== module.js ===" && ls -lah dist/module.js
echo "=== all JS chunks ===" && ls -lah dist/*.js | sort -k5 -rh | head -20
echo "=== chunk count ===" && ls dist/*.js | wc -l

Record the baseline. A pre-split plugin commonly has a module.js of 1–3 MB with no other JS chunks.


Step 3: Check and update create-plugin

The @grafana/create-plugin tool controls .config/webpack/, .config/jest/, and other build scaffolding. Updating it often unlocks faster SWC compilation and better chunk output.

bash
cat .config/.cprc.json 2>/dev/null || grep '"@grafana/create-plugin"' package.json
npm view @grafana/create-plugin version
npx @grafana/create-plugin@latest update

After updating, review the diff (especially .config/webpack/webpack.config.ts) and run a test build. If the plugin has a top-level webpack.config.ts that webpack-merges the base config, review the merge for conflicts.


Step 4: Analyse the codebase — find what to split

Do not start implementing until you have read all of these.

bash
# Entry point — look for direct (non-lazy) imports of App, ConfigPage, exposeComponent targets
cat src/module.ts 2>/dev/null || cat src/module.tsx

# Root App component — look for direct page/route imports that should be lazy
cat src/App.tsx src/components/App.tsx src/feature/app/components/App.tsx 2>/dev/null | head -80

# Extension registrations — each should become an independent chunk
grep -r "exposeComponent\|addComponent\|addLink" src/ --include="*.ts" --include="*.tsx" -n

# Exported side-effect singletons (Faro, analytics) — must be extracted before splitting
grep -n "^export const\|^export let" src/module.ts src/module.tsx 2>/dev/null
grep -rn "from '.*module'" src/ --include="*.ts" --include="*.tsx" | grep -v node_modules

# Heavy synchronous imports
grep -rn "from 'monaco-editor\|@codemirror\|d3\b\|recharts\|chart\.js" \
  src/ --include="*.ts" --include="*.tsx" | grep -v node_modules

Key rule: If a file is imported by module.ts directly (even transitively), it ends up in module.js. Everything reachable from a lazy boundary becomes its own chunk.


Step 5: Implement splits — in priority order

Named vs default exports: React.lazy() requires a default export. Most Grafana plugin components use named exports — use .then() to re-map:

ts
// Named export
const LazyMyComp = lazy(() => import('./MyComponent').then(m => ({ default: m.MyComponent })));
// Default export
const LazyMyComp = lazy(() => import('./MyComponent'));

Priority 1: module.tsx (highest impact, always do this first)

If the entry point is module.ts, rename it: git mv src/module.ts src/module.tsx

Make module.tsx import nothing from feature code except through lazy():

tsx
import React, { lazy, Suspense } from 'react';
import { AppPlugin, AppRootProps } from '@grafana/data';
import { LoadingPlaceholder } from '@grafana/ui';

import type { MyExtensionProps } from './extensions/MyExtension';  // import type — erased at compile time
import type { JsonData } from './features/app/state/slice';

// Lazy Faro init — keeps @grafana/faro-react out of module.js
let faroInitialized = false;
async function initFaro() {
  if (faroInitialized) { return; }
  faroInitialized = true;
  const { initializeFaro } = await import('faro');
  initializeFaro();
}

const LazyApp = lazy(async () => {
  await initFaro();
  return import('./features/app/App').then(m => ({ default: m.App }));
});

function App(props: AppRootProps<JsonData>) {
  return <Suspense fallback={<LoadingPlaceholder text="" />}><LazyApp {...props} /></Suspense>;
}

const LazyMyExtension = lazy(() =>
  import('./extensions/MyExtension').then(m => ({ default: m.MyExtension }))
);
function MyExtension(props: MyExtensionProps) {
  return <Suspense fallback={<LoadingPlaceholder text="" />}><LazyMyExtension {...props} /></Suspense>;
}

export const plugin = new AppPlugin<JsonData>().setRootPage(App);
plugin.exposeComponent({ id: 'my-plugin/my-extension/v1', title: 'My Extension', component: MyExtension });

Key details:

  • import type for props prevents webpack from following the import into the eager bundle
  • Use new AppPlugin<JsonData>() if App uses AppRootProps<JsonData> — without the generic, setRootPage() type won't match
  • Remove any App as unknown as ComponentClass<AppRootProps> cast — the lazy wrapper is a valid function component

Expected impact: module.js drops from MB range to ~50–200 KB.

Singletons (e.g. Faro): If module.ts has export const faro = initializeFaro(), do NOT keep it as a top-level import. Extract it to src/faro.ts, update all internal imports from '*/module' → '*/faro', then use the dynamic initFaro() pattern above.


Priority 2: Route-based splitting in App.tsx

tsx
import React, { lazy, Suspense } from 'react';
import { Route, Routes } from 'react-router-dom';
import { LoadingPlaceholder } from '@grafana/ui';

const HomePage     = lazy(() => import('../pages/Home'));
const SettingsPage = lazy(() => import('../pages/Settings'));
const DetailPage   = lazy(() => import('../pages/Detail'));

function App(props: AppRootProps) {
  return (
    <Suspense fallback={<LoadingPlaceholder text="" />}>
      <Routes>
        <Route path="home"       element={<HomePage />} />
        <Route path="settings"   element={<SettingsPage />} />
        <Route path="detail/:id" element={<DetailPage />} />
        <Route path=""           element={<HomePage />} />
      </Routes>
    </Suspense>
  );
}
export default App;

Bypass barrel files: Target the actual component file in the import(), not an index.ts barrel that re-exports multiple things:

tsx
// Risky — barrel may pull in other heavy modules
const Catalog = lazy(() => import('features/catalog'));
// Better — only pulls in Catalog's tree
const Catalog = lazy(() => import('features/catalog/Catalog').then(m => ({ default: m.Catalog })));

Priority 3: Extension components

Each extension should export default its component. Use fallback={null} for extensions that load quickly:

tsx
// src/extensions/MyExtension.tsx
export default function MyExtension(props: MyExtensionProps) {
  return <AppProviders><MyExtensionContent {...props} /></AppProviders>;
}

Surgical split: If an extension wrapper must stay eager in module.tsx, lazy-load the heavy component it renders:

tsx
const HeavyInner = lazy(() => import('components/features/HeavyInner'));
export function MyExtension() {
  return <Suspense fallback={<LoadingPlaceholder text="" />}><HeavyInner /></Suspense>;
}

Priority 4: Component registries and tab panels

For arrays of objects containing React components (e.g. tab panels), lazy-load each entry. Critical: ensure a <Suspense> boundary exists where the component renders.

tsx
const ConfigDetails = lazy(() => import('./ConfigDetails/ConfigDetails').then(m => ({ default: m.ConfigDetails })));
const Overview      = lazy(() => import('./Overview/Overview').then(m => ({ default: m.Overview })));

const tabs = [
  { id: 'overview', component: Overview },
  { id: 'config',   component: ConfigDetails },
];

// In the parent that renders the active tab:
<Suspense fallback={<LoadingPlaceholder text="" />}>
  {ActiveTab && <ActiveTab />}
</Suspense>

For datasource plugins (setConfigEditor, setQueryEditor, VariableSupport, AnnotationSupport), see references/datasource-plugins.md.


Step 6: Group related chunks if over-splitting

If the build produces more than ~25 JS files, use webpack magic comments:

tsx
const FleetList   = lazy(() => import(/* webpackChunkName: "fleet" */ '../pages/FleetList'));
const FleetDetail = lazy(() => import(/* webpackChunkName: "fleet" */ '../pages/FleetDetail'));

One webpackChunkName per logical feature area. Don't group unrelated pages.


Step 7: Measure and verify

bash
yarn build 2>/dev/null || npm run build
echo "=== module.js ===" && ls -lah dist/module.js
echo "=== all JS chunks (largest first) ===" && ls -lah dist/*.js | sort -k5 -rh | head -30
echo "=== chunk count ===" && ls dist/*.js | wc -l
MetricTarget
module.js size< 200 KB
Total JS chunk count15–25
Largest single chunk< 1 MB
bash
# Analyse bundle composition if a chunk is unexpectedly large
npx webpack-bundle-analyzer dist/stats.json 2>/dev/null

Step 8: Test the running plugin

  1. Open the plugin in a Grafana instance
  2. Navigate to every route — each triggers a new chunk download
  3. DevTools → Network → JS: confirm lazy chunks load on navigation, not all upfront
  4. Check Console for errors
  5. Test any exposeComponent extensions from other Grafana apps

For troubleshooting common issues, see references/troubleshooting.md.


References

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Plugin Bundle Size AI skill do?

Optimise Grafana app plugin bundle size using React.lazy, Suspense, and webpack code splitting. Use when the user asks to reduce plugin bundle size, optimise module.js, add code splitting, improve initial plugin load performance, split plugin chunks, lazy load plugin pages, or help implement lazy loading in a Grafana app plugin. Triggers on phrases like "optimise plugin bundle size", "module.js is too large", "plugin is slow to load", "code split the plugin", "reduce initial JS payload", or "help me with Suspense in my plugin".

Why use Plugin Bundle Size on TypingMind?

Because you install it once and use it with any model. Plugin Bundle Size is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Plugin Bundle Size in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/grafana/skills/tree/main/skills/grafana-plugins/plugin-bundle-size. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Plugin Bundle Size?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Plugin Bundle Size?

As many as you like. As long as a model supports skills, you can use Plugin Bundle Size with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Plugin Bundle Size AI skill free?

Yes. It is published on GitHub by grafana under the Apache-2.0 license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇