vue-renderer-markdown

Fast, streaming-friendly Markdown rendering for Vue 3 — progressive Mermaid, streaming diff code blocks, and real-time previews optimized for large documents.

Table of Contents

• Why use it?
• Compared to traditional Markdown renderers
• Live Demo
• Features
• Install
  • Peer Dependencies
• Server-Side Rendering (SSR)
• Math rendering options
• Quick Start
  • Choose Your Code Block Style
• TypeScript Usage
• Why vue-renderer-markdown?
• Usage
• Performance Features
• Performance Tips
  • Props
• New prop: renderCodeBlocksAsPre
• Advanced customization
• Monaco Editor Integration
• Code block header customization
• Mermaid: Progressive Rendering Example
• Tailwind (e.g. shadcn) — fix style ordering issues
• Troubleshooting
• Thanks
• Star History
• License

Why use it?

• Progressive Mermaid: diagrams render incrementally so users see results earlier.
• Streaming diff code blocks: show diffs as they arrive for instant feedback.
• Built for scale: optimized DOM updates and memory usage for very large documents.

Compared to traditional Markdown renderers

Traditional Markdown renderers typically convert a finished Markdown string into a static HTML tree. This library is designed for streaming and interactive workflows and therefore provides capabilities you won't find in a classic renderer:

• Streaming-first rendering: render partial or incrementally-updated Markdown content without re-parsing the whole document each time. This enables live previews for AI outputs or editors that emit tokens progressively.
• Streaming-aware code blocks and "code-jump" UX: large code blocks are updated incrementally and the renderer can maintain cursor/selection context and fine-grained edits. This enables smooth code-editing experiences and programmatic "jump to" behaviors that traditional renderers do not support.
• Built-in diff/code-stream components: show diffs as they arrive (line-by-line or token-by-token) with minimal reflow. This is ideal for streaming AI edits or progressive code reviews — functionality that is not available in plain Markdown renderers.
• Progressive diagrams and editors: Mermaid diagrams and Monaco-based previews update progressively and render as soon as they become valid.
• Flexible code block rendering: Choose between full Monaco Editor integration for interactive editing or lightweight Shiki-based syntax highlighting for display-only scenarios.
• Smooth, interactive UI: the renderer is optimized for minimal DOM churn and silky interactions (e.g. streaming diffs, incremental diagram updates, and editor integrations) so UX remains responsive even with very large documents.

These features make the library especially suited for real-time, AI-driven, and large-document scenarios where a conventional, static Markdown-to-HTML conversion would lag or break the user experience.

🚀 Live Demo

• Streaming playground — try large Markdown files and progressive diagrams to feel the difference.
• Markdown vs v-html comparison — contrast the library's reactive rendering with a traditional static pipeline.

Features

• ⚡ Ultra-High Performance: Optimized for real-time streaming with minimal re-renders and efficient DOM updates
• 🌊 Streaming-First Design: Built specifically to handle incomplete, rapidly updating, and tokenized Markdown content
• 🧠 Monaco Streaming Updates: High-performance Monaco integration with smooth, incremental updates for large code blocks
• 🪄 Progressive Mermaid Rendering: Diagrams render as they become valid and update incrementally without jank
• 🧩 Custom Components: Seamlessly integrate your Vue components within Markdown content
• 📝 Complete Markdown Support: Tables, math formulas, emoji, checkboxes, code blocks, and more
• 🔄 Real-Time Updates: Handles partial content and incremental updates without breaking formatting
• 📦 TypeScript First: Full type definitions with intelligent auto-completion
• 🔌 Zero Configuration: Drop-in component that works with any Vue 3 project out of the box
• 🎨 Flexible Code Rendering: Choose between Monaco Editor integration (CodeBlockNode) or lightweight markdown-style syntax highlighting (MarkdownCodeBlockNode)

Install

pnpm add vue-renderer-markdown
# or
npm install vue-renderer-markdown
# or
yarn add vue-renderer-markdown

Peer Dependencies

Custom parse hooks

For advanced use-cases you can inject hooks into the parsing pipeline via parseMarkdownToStructure's options argument or by calling the utility directly when you need an AST.

• preTransformTokens?: (tokens: MarkdownToken[]) => MarkdownToken[] — called immediately after markdown-it produces tokens. Use this to rewrite or replace tokens before library parsing.
• postTransformTokens?: (tokens: MarkdownToken[]) => MarkdownToken[] — called after internal fixes; if you return a different token array instance it will be re-processed into nodes.
• postTransformNodes?: (nodes: ParsedNode[]) => ParsedNode[] — operate directly on the parsed node tree. This is often the simplest and most efficient way to adjust the final output.

Token-level example (preTransformTokens):

import { getMarkdown, parseMarkdownToStructure } from 'vue-renderer-markdown'

const md = getMarkdown()

function pre(tokens) {
  return tokens.map((t) => {
    if (t.type === 'html_block' && /<thinking>/.test(t.content || '')) {
      return { ...t, type: 'thinking_block', content: (t.content || '').replace(/<\/?thinking>/g, '') }
    }
    return t
  })
}

const nodes = parseMarkdownToStructure(markdownString, md, { preTransformTokens: pre })

Node-level example (postTransformNodes):

function postNodes(nodes) {
  if (!nodes || nodes.length === 0)
    return nodes
  const first = nodes[0]
  if (first.type === 'paragraph') {
    return [{ type: 'thinking', content: 'Auto-thought', children: [first] }, ...nodes.slice(1)]
  }
  return nodes
}

const nodes2 = parseMarkdownToStructure(markdownString, md, { postTransformNodes: postNodes })

Playground demo

The included playground demonstrates a small, scoped custom component mapping example: playground/src/components/ThinkingNode.vue renders type: 'thinking' nodes. The playground registers this mapping with setCustomComponents('playground-demo', { thinking: ThinkingNode }) and the MarkdownRender instance in playground/src/pages/index.vue uses custom-id="playground-demo". You can toggle the demo at runtime in the playground's settings panel.

This package requires Vue 3. Math (LaTeX) rendering is optional and requires installing katex as a peer dependency — KaTeX is not bundled or auto-injected. Additional optional peer dependencies enable advanced features and are lazy-loaded at runtime when available.

Required Peer Dependencies

Vue 3 (required for all features):

# pnpm (recommended)
pnpm add vue

# npm
npm install vue

# yarn
yarn add vue

Optional Peer Dependencies

Install these to enable advanced features. The library will gracefully degrade if they are not available.

Full install (recommended if you want all features):

# pnpm
pnpm add mermaid stream-monaco shiki

# npm
npm install mermaid stream-monaco shiki

# yarn
yarn add mermaid stream-monaco shiki

Individual optional features:

Peer Dependency	Version	Enables	Fallback if missing
mermaid	>=11	Progressive Mermaid diagram rendering	Shows code block source
stream-monaco	>=0.0.33	Monaco Editor for interactive code editing	Plain text display
stream-markdown	>=0.0.2	Streaming renderer used by MarkdownCodeBlockNode	Plain text display
shiki	^3.13.0	Syntax highlighting for MarkdownCodeBlockNode	Plain text display
vue-i18n	>=9	Internationalization support	Built-in fallback translator

Important Notes:

⚠️ KaTeX is not bundled or auto-injected. To enable LaTeX math rendering install katex and import its CSS in your application. Example:

pnpm add katex
# or
npm install katex

Then import the stylesheet in your app entry (for example main.ts):

import 'katex/dist/katex.min.css'

🖼️ Toolbar icons ship as local SVGs—no additional icon libraries required

• The exact peer version ranges are declared in this package's package.json
• Optional peers are lazy-loaded at runtime, so you can start with minimal dependencies and add features later
• For monorepos or pnpm workspaces, install peers at the workspace root to ensure they are available to consuming packages

Server-Side Rendering (SSR)

This library is designed to be import-safe in SSR builds. Heavy dependencies (Monaco, Mermaid) are lazy-loaded at runtime and browser-only features are properly guarded. However, some advanced features (Monaco editor, progressive Mermaid rendering, Web Workers) require browser APIs and must be rendered client-side only.

Quick Start: Nuxt 3

Use Nuxt's <client-only> wrapper:

<template>
  <client-only>
    <MarkdownRender :content="markdown" />
  </client-only>
</template>

For detailed Nuxt 3 setup, see: docs/nuxt-ssr.md

Vite SSR / Custom SSR

Use a client-only wrapper with Vue lifecycle hooks:

<script setup lang="ts">
import { onMounted, ref } from 'vue'
import MarkdownRender from 'vue-renderer-markdown'

const mounted = ref(false)
onMounted(() => {
  mounted.value = true
})
</script>

<template>
  <div v-if="mounted">
    <MarkdownRender :content="markdown" />
  </div>
  <div v-else>
    <!-- SSR fallback: lightweight preformatted text -->
    <pre>{{ markdown }}</pre>
  </div>
</template>

SSR Testing

Run the SSR smoke test to verify import safety:

pnpm run check:ssr

This test ensures the library can be imported in a Node environment without throwing errors.

Notes

• Optional peers (Mermaid, Monaco, icons) are lazy-loaded only in the browser -- Math rendering (KaTeX) works during SSR when katex is installed and available to the build. Since KaTeX is not bundled, ensure you add katex as a dependency if you need server-side math rendering.
• For pre-rendered diagrams/code, generate static HTML server-side and pass it as raw HTML or AST
• If you encounter window is not defined errors, please open an issue with the stack trace

Math rendering options

This library includes a lightweight math inline/block plugin that attempts to normalize common KaTeX/TeX commands and accidental control characters (for example when "\b" was interpreted as a backspace character by JS).

You can customize the behavior via getMarkdown's mathOptions parameter:

import { getMarkdown } from './src/utils/markdown/getMarkdown'

const md = getMarkdown({
  mathOptions: {
    // override which words should be auto-prefixed with a backslash
    commands: ['in', 'perp', 'alpha'],
    // whether to escape standalone '!' (default: true)
    escapeExclamation: true,
  }
})

There are also two exported helpers you can use directly:

• KATEX_COMMANDS — default list of command words the plugin will auto-escape when missing a leading \.
• normalizeStandaloneBackslashT(s, opts?) — the normalization helper used internally. You can call it yourself if you need to pre-process math content before handing it to KaTeX.

Example:

import { KATEX_COMMANDS, normalizeStandaloneBackslashT } from 'vue-renderer-markdown'

const raw = 'a\tb + infty'
const normalized = normalizeStandaloneBackslashT(raw, { commands: KATEX_COMMANDS })
// normalized is now safe to pass to KaTeX

Plugin install example (global defaults)

You can set global math options when installing the Vue plugin so all markdown instances created by the library inherit the same defaults.

import { createApp } from 'vue'
import MarkdownRender, { VueRendererMarkdown } from 'vue-renderer-markdown'

const app = createApp(App)

// Set global math options during plugin install
app.use(VueRendererMarkdown, {
  mathOptions: {
    commands: ['in', 'perp', 'alpha'],
    escapeExclamation: false,
  }
})

app.mount('#app')

Alternatively, you can programmatically set the global defaults by importing setDefaultMathOptions:

import { setDefaultMathOptions } from 'vue-renderer-markdown'

setDefaultMathOptions({ commands: ['infty', 'perp'], escapeExclamation: true })

Internationalization (i18n)

By default, getMarkdown uses English text for UI elements (e.g., "Copy" button in code blocks). You can customize these texts by providing an i18n option:

Using a translation map:

import { getMarkdown } from 'vue-renderer-markdown'

const md = getMarkdown('editor-1', {
  i18n: {
    'common.copy': '复制',
  }
})

Using a translation function:

import { useI18n } from 'vue-i18n' // or any i18n library
import { getMarkdown } from 'vue-renderer-markdown'

const { t } = useI18n()

const md = getMarkdown('editor-1', {
  i18n: (key: string) => t(key)
})

Default translations:

• common.copy: "Copy" — Used in code block copy buttons

This design keeps the markdown utilities pure and free from global side effects, allowing you to integrate with any i18n solution or provide static translations.

Quick Start

1. Install

pnpm add vue-renderer-markdown vue
# or
npm install vue-renderer-markdown vue
# or
yarn add vue-renderer-markdown vue

2. Basic Usage

<script setup lang="ts">
import MarkdownRender from 'vue-renderer-markdown'
import 'vue-renderer-markdown/index.css'

const content = `
# Hello World

This is **bold** and this is *italic*.

- List item 1
- List item 2

\`\`\`javascript
console.log('Code block!')
\`\`\`
`
</script>

<template>
  <MarkdownRender :content="content" />
</template>

3. Enable Optional Features

Mermaid Diagrams:

### NodeRenderer prop: `parseOptions`

The `<MarkdownRender />` component (a.k.a. `NodeRenderer`) now accepts a `parseOptions` prop which is forwarded to the internal `parseMarkdownToStructure` call when you pass `content` to the component. This lets you inject token- or node-level transforms from the component usage without calling the parser yourself.

Type shape (re-exported from the library):

- `preTransformTokens?: (tokens: MarkdownToken[]) => MarkdownToken[]` — called immediately after `markdown-it` produces tokens. Use to rewrite or replace tokens before the library processes them.
- `postTransformTokens?: (tokens: MarkdownToken[]) => MarkdownToken[]` — called after internal token fixes; if you return a different array it will be re-processed into nodes.
- `postTransformNodes?: (nodes: ParsedNode[]) => ParsedNode[]` — operate directly on the parsed node tree.

Why use the prop? Passing `parseOptions` is convenient when you want to support custom inline syntax or lightweight HTML-like tokens that should be mapped to custom node types and rendered with your own Vue components. Combine `parseOptions` with `setCustomComponents` (or the `custom-id` / per-instance custom components mechanism) to map custom node `type` values to Vue components.

Token-level example (pass as component prop):

```vue
<script setup lang="ts">
import MarkdownRender, { getMarkdown } from 'vue-renderer-markdown'

const md = getMarkdown()

function pre(tokens: any[]) {
  return tokens.map((t) => {
    if (t.type === 'html_block' && /<thinking>/.test(t.content || '')) {
      return { ...t, type: 'thinking_block', content: (t.content || '').replace(/<\/?thinking>/g, '') }
    }
    return t
  })
}

const parseOptions = { preTransformTokens: pre }
</script>

<template>
  <MarkdownRender :content="markdownString" :parseOptions="parseOptions" custom-id="playground-demo" />
</template>

Node-level example (postTransformNodes as a component prop):

<script setup lang="ts">
import MarkdownRender from 'vue-renderer-markdown'

function postNodes(nodes) {
  if (!nodes || nodes.length === 0)
    return nodes
  const first = nodes[0]
  if (first.type === 'paragraph') {
    return [{ type: 'thinking', content: 'Auto-thought', children: [first] }, ...nodes.slice(1)]
  }
  return nodes
}

const parseOptions = { postTransformNodes: postNodes }
</script>

<template>
  <MarkdownRender :content="markdownString" :parse-options="parseOptions" />
</template>

Notes:

• When you create custom node types via token transforms, register the corresponding Vue component with setCustomComponents('your-id', { your_node_type: YourComponent }) and pass custom-id="your-id" to the MarkdownRender instance so it can look up and render your component.
• If you already call parseMarkdownToStructure yourself and pass nodes to the component, parseOptions is ignored — it only applies when content is provided and the component does the parsing.

pnpm add mermaid

**Monaco Editor (Interactive Code Editing):**
```bash
pnpm add stream-monaco
# Also configure vite-plugin-monaco-editor-esm (see Monaco section)

Syntax Highlighting (Lightweight Alternative to Monaco):

pnpm add shiki

Choose Your Code Block Style

The library offers flexible code block rendering:

Mode	Component	Best for	Dependencies
Monaco Editor (default)	CodeBlockNode	Rich editing, streaming diffs, toolbar actions	stream-monaco
Shiki Syntax Highlighting	MarkdownCodeBlockNode	Lightweight read-only views, SSR friendly output	shiki
Plain Text	PreCodeNode	Minimal dependencies, AI reasoning traces, logging output	None

Default: Monaco Editor Integration (full-featured)

• Interactive editing
• Advanced features (copy, expand, preview)
• Requires stream-monaco peer dependency

Alternative: Shiki Syntax Highlighting (lightweight)

<script setup lang="ts">
import { MarkdownCodeBlockNode, setCustomComponents } from 'vue-renderer-markdown'

// Override globally to use lightweight rendering
setCustomComponents({
  code_block: MarkdownCodeBlockNode,
})
</script>

Minimal: Plain Text (no dependencies)

<MarkdownRender :content="content" :render-code-blocks-as-pre="true" />

That's it! See the sections below for advanced features and customization.

TypeScript Usage

Typed AST rendering

<script setup lang="ts">
import type { BaseNode } from 'vue-renderer-markdown'
import { ref, watchEffect } from 'vue'
import MarkdownRender, { parseMarkdownToStructure } from 'vue-renderer-markdown'

const content = ref<string>('# Hello \n\n```ts\nconsole.log(1)\n```')
const nodes = ref<BaseNode[]>([])

watchEffect(() => {
  nodes.value = parseMarkdownToStructure(content.value)
})
</script>

<template>
  <MarkdownRender :nodes="nodes" />
</template>

Strongly typed custom components

<!-- components/CustomCodeBlock.vue -->
<script setup lang="ts">
import type { CodeBlockNode } from 'vue-renderer-markdown'

const props = defineProps<{ node: CodeBlockNode }>()
</script>

<template>
  <pre class="custom-code">
    <code :data-lang="props.node.language">{{ props.node.code }}</code>
  </pre>
</template>

// main.ts
import { createApp } from 'vue'
import { setCustomComponents, VueRendererMarkdown } from 'vue-renderer-markdown'
import App from './App.vue'
import CustomCodeBlock from './components/CustomCodeBlock.vue'

const app = createApp(App)

setCustomComponents('docs', {
  code_block: CustomCodeBlock,
})

app.use(VueRendererMarkdown, {
  mathOptions: {
    commands: ['infty', 'perp', 'alpha'],
    escapeExclamation: true,
  },
  getLanguageIcon(lang) {
    return lang === 'shell' ? '<span>sh</span>' : undefined
  },
})

app.mount('#app')

Why vue-renderer-markdown?

Streaming Markdown content from AI models, live editors, or real-time updates presents unique challenges:

• Incomplete syntax blocks can break traditional parsers
• Rapid content changes cause excessive re-renders and performance issues
• Cursor positioning becomes complex with dynamic content
• Partial tokens need graceful handling without visual glitches

vue-renderer-markdown solves these challenges with a streaming-optimized architecture that maintains perfect formatting and performance, even with the most demanding real-time scenarios.

Usage

Streaming Markdown (Recommended)

Perfect for AI model responses, live content updates, or any scenario requiring real-time Markdown rendering:

<script setup lang="ts">
import { ref } from 'vue'
import MarkdownRender from 'vue-renderer-markdown'

const content = ref('')
const fullContent = `# Streaming Content\n\nThis text appears character by character...`

// Simulate streaming content
let index = 0
const interval = setInterval(() => {
  if (index < fullContent.length) {
    content.value += fullContent[index]
    index++
  }
  else {
    clearInterval(interval)
  }
}, 50)
</script>

<template>
  <MarkdownRender :content="content" />
</template>

Basic Usage

For static or pre-generated Markdown content:

<script setup lang="ts">
import MarkdownRender from 'vue-renderer-markdown'

const markdownContent = `
# Hello Vue Markdown

This is **markdown** rendered as HTML!

- Supports lists
- [x] Checkboxes
- :smile: Emoji
`
</script>

<template>
  <MarkdownRender :content="markdownContent" />
</template>

Performance Features

The streaming-optimized engine delivers:

• Incremental Parsing Code Blocks: Only processes changed content, not the entire code block
• Efficient DOM Updates: Minimal re-renders
• Monaco Streaming: Fast, incremental updates for large code snippets without blocking the UI
• Progressive Mermaid: Diagrams render as soon as syntax is valid and refine as content streams in
• Memory Optimized: Intelligent cleanup prevents memory leaks during long streaming sessions
• Animation Frame Based: Smooth animations
• Graceful Degradation: Handles malformed or incomplete Markdown without breaking

Performance Tips

• Stream long documents in chunks to avoid blocking the main thread; the renderer incrementally patches the DOM.
• Prefer MarkdownCodeBlockNode or render-code-blocks-as-pre when you only need read-only output — this skips Monaco initialization.
• Scope custom component overrides with setCustomComponents(id, mapping) so unused components can be garbage-collected.
• Use the built-in setDefaultMathOptions helper once during app bootstrap to avoid repeatedly computing math config per render.
• When Mermaid diagrams are heavy, pre-validate or pre-render them server-side and feed the resulting HTML as cached content.

Props

Name	Type	Required	Description
content	string	✓	Markdown string to render
nodes	BaseNode[]		Parsed markdown AST nodes (alternative to content)
renderCodeBlocksAsPre	boolean		When true, render all code_block nodes as simple <pre><code> blocks (uses PreCodeNode) instead of the full CodeBlockNode. Useful for lightweight, dependency-free rendering of multi-line text such as AI "thinking" outputs. Defaults to false.
codeBlockStream	boolean		Controls streaming behavior for code_block nodes. When true (default), code blocks update progressively as content streams in. When false, code blocks stay in a lightweight loading state and render only once when the final code is ready (defers Monaco creation).
viewportPriority	boolean		When enabled (default), heavy nodes (e.g. Mermaid, Monaco) prioritize rendering for content within or near the viewport, deferring offscreen work to improve responsiveness. Set to false to render everything eagerly (useful for print/export or when you need immediate layout for all nodes). Defaults to true.

Either content or nodes must be provided.

Note: when using the component in a Vue template, camelCase prop names should be written in kebab-case (for example, renderCodeBlocksAsPre -> render-code-blocks-as-pre).

New prop: renderCodeBlocksAsPre

• Type: boolean
• Default: false

Description:

• When set to true, all parsed code_block nodes are rendered as a simple <pre><code> (the library's internal PreCodeNode) instead of the full CodeBlockNode which may depend on optional peers such as Monaco or mermaid.
• Use case: enable this when you need lightweight, preformatted text rendering (for example AI "thinking" outputs or multi-line reasoning steps) and want to avoid depending on optional peer libraries while preserving original formatting.

Notes:

• When renderCodeBlocksAsPre: true, props passed to CodeBlockNode such as codeBlockDarkTheme, codeBlockMonacoOptions, themes, minWidth, maxWidth, etc. will not take effect because CodeBlockNode is not used.
• If you need the full code block feature set (syntax highlighting, folding, copy button, etc.), keep the default false and install the optional peers (mermaid, stream-monaco).

Example (Vue usage):

<script setup lang="ts">
import MarkdownRender from 'vue-renderer-markdown'

const markdown = `Here is an AI thinking output:\n\n\`\`\`text\nStep 1...\nStep 2...\n\`\`\`\n`
</script>

<template>
  <MarkdownRender :content="markdown" :render-code-blocks-as-pre="true" />
</template>

New prop: codeBlockStream

• Type: boolean
• Default: true

Description:

• When set to false, code_block nodes won't stream intermediate updates. Instead, they remain in a loading state and render once when the final code is available. This can reduce layout churn and avoid initializing Monaco repeatedly for partially complete content, which may improve performance in certain scenarios.

Notes:

• This option only applies when CodeBlockNode is used (i.e., renderCodeBlocksAsPre is false).
• While loading, the component shows a lightweight placeholder to keep layout stable. Monaco editor initialization is deferred until loading becomes false.

Example (Vue usage):

<MarkdownRender :content="markdown" :code-block-stream="false" />

New prop: viewportPriority

• Type: boolean
• Default: true

Description:

• When enabled (default), the renderer defers expensive work for offscreen nodes and prioritizes rendering for elements in or near the viewport. This reduces time-to-interactive for long documents and streaming content, especially with heavy components like Mermaid and Monaco.
• Set to false to render all nodes eagerly. This can be helpful for print/export, pre-measuring layouts offscreen, or scenarios where you explicitly want all content rendered immediately.

Example:

<script setup lang="ts">
import MarkdownRender from 'vue-renderer-markdown'

const markdown = `# Long document with many diagrams and code blocks...`
</script>

<template>
  <!-- Disable prioritized/lazy rendering when preparing a printable/exportable view -->
  <MarkdownRender :content="markdown" :viewport-priority="false" />
  <!-- Note: in templates, use kebab-case: viewport-priority -->
  <!-- Default behavior (omit the prop) keeps it enabled -->
</template>

Advanced

Custom Components

You can override how internal node types are rendered by supplying a mapping from node keys to your Vue components. This library supports two approaches:

• Scoped per-instance mappings (recommended): provide a customId prop to MarkdownRender and call setCustomComponents(id, mapping) to scope overrides to that renderer instance.
• Legacy global mapping: call setCustomComponents(mapping) with a single argument. This remains supported for backward compatibility but is less flexible and is considered deprecated in new code.

Scoped example (recommended):

import { createApp } from 'vue'
import MarkdownRender, { setCustomComponents } from 'vue-renderer-markdown'
import App from './App.vue'
import MyCustomNode from './components/MyCustomNode.vue'

const app = createApp(App)

// Scope this mapping to instances that use customId="docs-page"
setCustomComponents('docs-page', {
  admonition: MyCustomNode,
  // ...other overrides
})

app.mount('#app')

Then, pass the matching customId prop to the MarkdownRender instance you want to affect:

<MarkdownRender :content="markdownContent" custom-id="docs-page" />

If you create scoped mappings dynamically (for example in a single-page app that mounts/unmounts multiple different renderers), you can remove a mapping to free memory or avoid stale overrides:

import { removeCustomComponents } from 'vue-renderer-markdown'

removeCustomComponents('docs-page')

Legacy/global example (backwards compatible):

// Deprecated-style global mapping (still supported)
setCustomComponents({
  code_block: MarkdownCodeBlockNode,
})

MarkdownCodeBlockNode: Alternative Code Block Renderer

The library now includes MarkdownCodeBlockNode - an alternative code block component that provides markdown-style syntax highlighting instead of Monaco Editor integration. This gives you the flexibility to choose between two rendering approaches for code blocks:

• CodeBlockNode (default): Full-featured code blocks with Monaco Editor integration, copy buttons, expand/collapse, and advanced features
• MarkdownCodeBlockNode: Lightweight markdown-style rendering with syntax highlighting using Shiki

Required peers when using MarkdownCodeBlockNode:

• stream-markdown (>=0.0.2)
• shiki (^3.13.0)

Install examples:

# pnpm
pnpm add stream-markdown shiki

# npm
npm install stream-markdown shiki

# yarn
yarn add stream-markdown shiki

When to use MarkdownCodeBlockNode:

• You want syntax-highlighted code blocks without Monaco Editor dependencies
• You prefer a lighter-weight solution for code display
• You need consistent markdown-style rendering across your application
• You don't need Monaco's editing capabilities

Usage Example:

import { createApp } from 'vue'
import MarkdownRender, { MarkdownCodeBlockNode, setCustomComponents } from 'vue-renderer-markdown'
import App from './App.vue'

const app = createApp(App)

// Override code_block to use markdown-style rendering
setCustomComponents({
  code_block: MarkdownCodeBlockNode,
})

app.mount('#app')

MarkdownCodeBlockNode Props:

Name	Type	Default	Description
node	CodeBlockNode	-	The code block node object
loading	boolean	true	Whether to show loading state
darkTheme	string	'vitesse-dark'	Dark theme for syntax highlighting
lightTheme	string	'vitesse-light'	Light theme for syntax highlighting
isDark	boolean	false	Whether to use dark theme
themes	string[]	-	Array of [darkTheme, lightTheme] for highlighting
showHeader	boolean	true	Whether to show the code block header

The component automatically handles Mermaid diagrams and provides clean syntax highlighting for all other languages using Shiki themes.

Notes:

Notes:

• Use the scoped API when you need different component mappings for different renderer instances — e.g. one mapping for a docs site and another for an editor preview. Call setCustomComponents('my-id', mapping) and pass custom-id="my-id" to the MarkdownRender instance.
• The single-argument form setCustomComponents(mapping) continues to work as a global fallback but is deprecated for new usage.
• When using MarkdownCodeBlockNode, Monaco Editor related props won't have any effect since that component uses Shiki for highlighting instead.
• When using MarkdownCodeBlockNode, Monaco Editor related props won't have any effect since this component uses Shiki for highlighting instead.

TypeScript: Full type support. Import types as needed:

import type { MyMarkdownProps } from 'vue-renderer-markdown/dist/types'

ImageNode slots (placeholder / error)

ImageNode now supports two named slots so you can customize the loading and error states:

• Slot name: placeholder
• Slot name: error

Both slots receive the same set of reactive slot props:

• node — the original ImageNode object ({ type: 'image', src, alt, title, raw })
• displaySrc — the current src used for rendering (will be fallbackSrc if a fallback was applied)
• imageLoaded — boolean, whether the image has finished loading
• hasError — boolean, whether the image is in an error state
• fallbackSrc — string, the fallback src passed to the component (if any)
• lazy — boolean, whether lazy loading is used
• isSvg — boolean, whether the current displaySrc is an SVG

Default behavior: if you don't provide the slots the component shows a built-in CSS spinner for placeholder and a simple error placeholder for error.

Example: customize loading and error slots

<ImageNode :node="node" :fallback-src="fallback" :lazy="true">
  <template #placeholder="{ node, displaySrc, imageLoaded }">
    <div class="p-4 bg-gray-50 rounded shadow-sm flex items-center justify-center">
      <div class="animate-pulse w-full h-24 bg-gray-200"></div>
      <span class="sr-only">Loading image</span>
    </div>
  </template>

  <template #error="{ node, displaySrc }">
    <div class="p-4 text-sm text-red-600 flex items-center gap-2">
      <strong>Failed to load image</strong>
      <span class="truncate">{{ displaySrc }}</span>
    </div>
  </template>
</ImageNode>

Tip: to avoid layout shift when switching from placeholder to the image, keep the placeholder's width/height similar to the final image (or use aspect-ratio / min-height). This lets the image fade/transform without triggering layout reflow.

TableNode loading slot

TableNode ships with a lightweight shimmer skeleton + spinner overlay that activates while node.loading is true. You can replace the overlay content without losing the shimmer effect by providing the named loading slot.

• Slot name: loading
• Slot props: { isLoading: boolean }

If you omit the slot the default spinner remains. The shimmer stays active either way because it is driven by the table cell CSS, so your custom slot can focus on messaging or branding.

Example: custom loading slot

<TableNode :node="node" index-key="demo">
  <template #loading="{ isLoading }">
    <div class="flex items-center gap-2 text-slate-500">
      <svg
        class="animate-spin h-5 w-5 text-slate-400"
        viewBox="0 0 24 24"
        aria-hidden="true"
      >
        <circle
          class="opacity-25"
          cx="12"
          cy="12"
          r="10"
          stroke="currentColor"
          stroke-width="4"
          fill="none"
        />
        <path
          class="opacity-75"
          fill="currentColor"
          d="M4 12a8 8 0 018-8v4a4 4 0 00-4 4H4z"
        />
      </svg>
      <span>
        {{ isLoading ? 'Fetching table rows…' : 'Loaded' }}
      </span>
    </div>
  </template>
</TableNode>

LinkNode: underline animation & color customization

LinkNode (the internal node used to render anchors) now supports runtime customization of underline animation and color via props — no need to override global CSS. Defaults preserve the previous appearance.

Available props (pass to the component that renders LinkNode):

Name	Type	Default	Description
color	string	#0366d6	Link text color (any valid CSS color). The underline uses currentColor, so it follows this color.
underlineHeight	number	2	Underline thickness in pixels.
underlineBottom	number | string	-3px	Offset from the text baseline; accepts px or any CSS length (e.g., 0.2rem).
animationDuration	number	0.8	Total animation duration in seconds.
animationOpacity	number	0.9	Underline opacity.
animationTiming	string	linear	CSS timing function (e.g., linear, ease, ease-in-out).
animationIteration	string | number	infinite	Animation iteration count or 'infinite'.

Example:

<template>
  <!-- Default styling -->
  <LinkNode :node="node" />

  <!-- Custom color and underline styling -->
  <LinkNode
    :node="node"
    color="#e11d48"
    :underline-height="3"
    underline-bottom="-4px"
    :animation-duration="1.2"
    :animation-opacity="0.8"
    animation-timing="ease-in-out"
  />
</template>

Notes:

• The underline color uses currentColor, so by default it matches the color prop. If you need an independent underline color, consider a small local CSS override or opening an issue to discuss exposing an underlineColor prop.
• All props are optional; when omitted, sensible defaults are used to preserve backward compatibility.

Override Language Icons

Override how code language icons are resolved via the plugin option getLanguageIcon. This keeps your usage unchanged and centralizes customization.

Plugin usage:

import { createApp } from 'vue'
import { VueRendererMarkdown } from 'vue-renderer-markdown'
import App from './App.vue'

const app = createApp(App)

// Example 1: replace shell/Shellscript icon with a remote SVG URL
const SHELL_ICON_URL = 'https://raw.githubusercontent.com/catppuccin/vscode-icons/refs/heads/main/icons/mocha/bash.svg'
app.use(VueRendererMarkdown, {
  getLanguageIcon(lang) {
    const l = (lang || '').toLowerCase()
    if (
      l === 'shell'
      || l === 'shellscript'
      || l === 'sh'
      || l === 'bash'
      || l === 'zsh'
      || l === 'powershell'
      || l === 'ps1'
      || l === 'bat'
      || l === 'batch'
    ) {
      return `<img src="${SHELL_ICON_URL}" alt="${l}" />`
    }
    // return empty/undefined to use the library default icon
    return undefined
  },
})

Local file example (import inline SVG):

import { createApp } from 'vue'
import { VueRendererMarkdown } from 'vue-renderer-markdown'
import App from './App.vue'
import JsIcon from './assets/javascript.svg?raw'

const app = createApp(App)

app.use(VueRendererMarkdown, {
  getLanguageIcon(lang) {
    const l = (lang || '').toLowerCase()
    if (l === 'javascript' || l === 'js')
      return JsIcon // inline SVG string
    return undefined
  },
})

Notes:

• The resolver returns raw HTML/SVG string. Returning undefined/empty value defers to the built-in mapping.
• Works across all code blocks without changing component usage.
• Alignment: icons render inside a fixed-size slot; both <svg> and <img> align consistently, no inline styles needed.
• For local files, import with ?raw and ensure the file is a pure SVG (not an HTML page). Download the raw SVG instead of GitHub’s HTML preview.
• The resolver receives the raw language string (e.g., tsx:src/components/file.tsx). The built-in fallback mapping uses only the base segment before :.

Monaco Editor Integration

If you are using Monaco Editor in your project, configure vite-plugin-monaco-editor-esm to handle global injection of workers. Our renderer is optimized for streaming updates to large code blocks—when content changes incrementally, only the necessary parts are updated for smooth, responsive rendering. On Windows, you may encounter issues during the build process. To resolve this, configure customDistPath to ensure successful packaging.

Note: If you only need to render a Monaco editor (for editing or previewing code) and don't require this library's full Markdown rendering pipeline, you can integrate Monaco directly using stream-monaco for a lighter, more direct integration.

pnpm add vite-plugin-monaco-editor-esm monaco-editor -d

npm equivalent:

npm install vite-plugin-monaco-editor-esm monaco-editor --save-dev

yarn equivalent:

yarn add vite-plugin-monaco-editor-esm monaco-editor -d

Example Configuration

import path from 'node:path'
import monacoEditorPlugin from 'vite-plugin-monaco-editor-esm'

export default {
  plugins: [
    monacoEditorPlugin({
      languageWorkers: [
        'editorWorkerService',
        'typescript',
        'css',
        'html',
        'json',
      ],
      customDistPath(root, buildOutDir, base) {
        return path.resolve(buildOutDir, 'monacoeditorwork')
      },
    }),
  ],
}

Tip: Preload Monaco workers for smoother first code-block rendering

If your application uses the Monaco editor for editable code blocks, you can call getUseMonaco() during app initialization or on page mount to proactively import stream-monaco and preload the required workers. This helps avoid a noticeable delay or flicker when the first code_block renders. Example:

// Call during app init or on page mount
import { getUseMonaco } from './src/components/CodeBlockNode/monaco'

// Trigger dynamic import + preload (failures are handled gracefully)
getUseMonaco()

Internally, getUseMonaco() will attempt to dynamically import stream-monaco and call the library's preload helper to register/load Monaco workers; if the module is unavailable (not installed or during SSR) the function returns null and the renderer falls back safely.

Internationalization / Fallback translations

If you don't want to install or use vue-i18n, the library ships with a small synchronous fallback translator used for common UI strings (copy, preview, image loading, etc.). You can replace the default English fallback map with your preferred language by calling setDefaultI18nMap at app startup:

import { setDefaultI18nMap } from 'vue-renderer-markdown'

setDefaultI18nMap({
  'common.copy': '复制',
  'common.copySuccess': '已复制',
  'common.decrease': '减少',
  'common.reset': '重置',
  'common.increase': '增加',
  'common.expand': '展开',
  'common.collapse': '折叠',
  'common.preview': '预览',
  'image.loadError': '图片加载失败',
  'image.loading': '正在加载图片...',
})

This is purely optional — if you do install vue-i18n, the library will prefer it at runtime and use the real translations provided by your i18n setup.

Code block header customization

The code block component now exposes a flexible header API so consumers can:

• Toggle the entire header on/off.
• Show or hide built-in toolbar buttons (copy, expand, preview, font-size controls).
• Fully replace the left or right header content via named slots.

This makes it easy to adapt the header to your application's UX or to inject custom controls.

Props (new)

Name	Type	Default	Description
showHeader	boolean	true	Toggle rendering of the header bar.
showCopyButton	boolean	true	Show the built-in copy button.
showExpandButton	boolean	true	Show the built-in expand/collapse button.
showPreviewButton	boolean	true	Show the built-in preview button (when preview is available).
showFontSizeButtons	boolean	true	Show the built-in font-size controls (also requires enableFontSizeControl).

Slots

• header-left — Replace the left side of the header (language icon + label by default).
• header-right — Replace the right side of the header (built-in action buttons by default).
• loading — Customize the loading placeholder shown when stream={false} and loading={true}. Receives slot props: { loading: boolean, stream: boolean }.

Example: hide the header

<CodeBlockNode
  :node="{ type: 'code_block', language: 'javascript', code: 'console.log(1)', raw: 'console.log(1)' }"
  :showHeader="false"
  :loading="false"
/>

Example: custom header via slots

<CodeBlockNode
  :node="{ type: 'code_block', language: 'html', code: '<div>Hello</div>', raw: '<div>Hello</div>' }"
  :loading="false"
  :showCopyButton="false"
>
  <template #header-left>
    <div class="flex items-center space-x-2">
      <!-- custom icon or label -->
      <span class="text-sm font-medium">My HTML</span>
    </div>
  </template>

  <template #header-right>
    <div class="flex items-center space-x-2">
      <button class="px-2 py-1 bg-blue-600 text-white rounded">Run</button>
      <button class="px-2 py-1 bg-gray-200 dark:bg-gray-700 rounded">Inspect</button>
    </div>
  </template>
</CodeBlockNode>

Example: custom loading placeholder

<CodeBlockNode
  :node="{ type: 'code_block', language: 'python', code: code, raw: code }"
  :stream="false"
  :loading="isLoading"
>
  <template #loading="{ loading, stream }">
    <div v-if="loading && !stream" class="p-4 text-center">
      <div class="animate-spin inline-block w-6 h-6 border-2 border-current border-t-transparent rounded-full" />
      <p class="mt-2 text-sm text-gray-500">Initializing editor...</p>
    </div>
  </template>
</CodeBlockNode>

Notes

• The new showFontSizeButtons prop provides an additional toggle; the existing enableFontSizeControl prop still controls whether the font-size feature is enabled at all. Keep both in mind when hiding/showing font controls.
• Existing behavior is unchanged by default — all new props default to true to preserve the original UI.

This configuration ensures that Monaco Editor workers are correctly packaged and accessible in your project.

Webpack — monaco-editor-webpack-plugin

If your project uses Webpack instead of Vite, you can use the official monaco-editor-webpack-plugin to bundle and inject Monaco's worker files. Here's a minimal example for Webpack 5:

Install:

# pnpm (dev)
pnpm add -D monaco-editor monaco-editor-webpack-plugin

# npm (dev)
npm install --save-dev monaco-editor monaco-editor-webpack-plugin

# yarn (dev)
yarn add -D monaco-editor monaco-editor-webpack-plugin

Note: pnpm add -D and yarn add -D are equivalent to npm install --save-dev and install the packages as development dependencies.

Example webpack.config.js:

const path = require('node:path')
const MonacoEditorPlugin = require('monaco-editor-webpack-plugin')

module.exports = {
  // ...your other config...
  output: {
    // Ensure worker files are placed correctly; adjust publicPath/filename as needed
    publicPath: '/',
  },
  plugins: [
    new MonacoEditorPlugin({
      // Limit to required languages/features to reduce bundle size
      languages: ['javascript', 'typescript', 'css', 'html', 'json'],
      // Optional: customize worker filename pattern
      filename: 'static/[name].worker.js',
    }),
  ],
}

Notes:

• For projects using monaco-editor, make sure the plugin handles the workers; otherwise the browser will try to load missing worker files at runtime (similar to Vite dep optimizer issues).
• If you see "file does not exist" errors after building (for example some workers are missing from the optimized deps directory), ensure the worker files are packaged into an accessible location via the plugin or build output.

Mermaid: Progressive Rendering Example

Mermaid diagrams can be streamed progressively. The diagram renders as soon as the syntax becomes valid and refines as more content arrives.

<script setup lang="ts">
import { ref } from 'vue'
import MarkdownRender from 'vue-renderer-markdown'

const content = ref('')
const steps = [
  '```mermaid\n',
  'graph TD\n',
  'A[Start]-->B{Is valid?}\n',
  'B -- Yes --> C[Render]\n',
  'B -- No  --> D[Wait]\n',
  '```\n',
]

let i = 0
const id = setInterval(() => {
  content.value += steps[i] || ''
  i++
  if (i >= steps.length)
    clearInterval(id)
}, 120)
</script>

<template>
  <MarkdownRender :content="content" />
  <!-- Diagram progressively appears as content streams in -->
  <!-- Mermaid must be installed as a peer dependency -->
</template>

Tailwind (e.g. shadcn) — fix style ordering issues

If your project uses a Tailwind component library like shadcn you may run into style ordering/override issues. We recommend importing the library CSS into a controlled Tailwind layer in your global stylesheet. For example, in your main stylesheet (e.g. src/styles/index.css or src/main.css):

/* main.css or index.css */
@tailwind base;
@tailwind components;
@tailwind utilities;

/* Recommended: place library styles into the components layer so your app components can override them */
@layer components {
  @import 'vue-renderer-markdown/index.css';
}

/* Alternative: place into the base layer if you want the library styles to be more foundational and harder to override:
@layer base {
  @import 'vue-renderer-markdown/index.css';
}
*/

Pick components (common) or base (when you want library styles to be more foundational) based on your desired override priority. After changing, run your dev/build command (e.g. pnpm dev) to verify the stylesheet ordering.

Troubleshooting

Monaco Editor workers not found

Symptom: Console errors like Could not load worker or Failed to load Monaco worker in production builds.

Solution: Configure vite-plugin-monaco-editor-esm in your vite.config.ts:

import path from 'node:path'
import monacoEditorPlugin from 'vite-plugin-monaco-editor-esm'

export default {
  plugins: [
    monacoEditorPlugin({
      languageWorkers: ['editorWorkerService', 'typescript', 'css', 'html', 'json'],
      customDistPath(root, buildOutDir, base) {
        return path.resolve(buildOutDir, 'monacoeditorwork')
      },
    }),
  ],
}

See the Monaco Editor Integration section for more details.

Mermaid diagrams not rendering

Symptom: Code blocks with mermaid language show plain text instead of rendered diagrams.

Solutions:

1. Install the mermaid peer dependency:

   pnpm add mermaid

2. Ensure your code block syntax is valid Mermaid:

   ```mermaid
   graph TD
     A[Start] --> B[End]
   ```

3. Check browser console for Mermaid errors. The library shows the source text if Mermaid rendering fails.

Syntax highlighting not working with MarkdownCodeBlockNode

Symptom: Code blocks show plain text without syntax highlighting when using MarkdownCodeBlockNode.

Solution: Install the required peers for MarkdownCodeBlockNode:

pnpm add stream-markdown shiki

TypeScript errors about missing types

Symptom: TypeScript errors like Cannot find module 'vue-renderer-markdown' or missing type definitions.

Solutions:

1. Import types from the correct path:

   import type { BaseNode, CodeBlockNode } from 'vue-renderer-markdown'

2. For specific type definitions:

   import type { MarkdownRenderProps } from 'vue-renderer-markdown/dist/types'

3. Ensure moduleResolution in tsconfig.json is set to "bundler" or "node16":

   {
     "compilerOptions": {
       "moduleResolution": "bundler"
     }
   }

Tailwind CSS styles conflicting

Symptom: Component styles are overridden by Tailwind utility classes or vice versa.

Solution: Import library CSS into a Tailwind layer in your main stylesheet:

/* main.css or index.css */
@tailwind base;
@tailwind components;
@tailwind utilities;

@layer components {
  @import 'vue-renderer-markdown/index.css';
}

See the Tailwind section for more details.

SSR: "window is not defined" errors

Symptom: Errors like ReferenceError: window is not defined during server-side rendering.

Solutions:

1. Nuxt 3: Wrap component in <client-only>:

   <template>
     <client-only>
       <MarkdownRender :content="markdown" />
     </client-only>
   </template>

2. Vite SSR: Use a client-only wrapper with lifecycle hooks:

   <script setup lang="ts">
   import { onMounted, ref } from 'vue'
   import MarkdownRender from 'vue-renderer-markdown'
   
   const mounted = ref(false)
   onMounted(() => {
     mounted.value = true
   })
   </script>
   
   <template>
     <div v-if="mounted">
       <MarkdownRender :content="markdown" />
     </div>
     <div v-else>
       <!-- SSR fallback: lightweight preformatted text -->
       <pre>{{ markdown }}</pre>
     </div>
   </template>

3. See the SSR section for complete setup guides.

Icons not showing

Symptom: Toolbar buttons show fallback elements instead of icons.

Solutions:

1. Ensure you are importing the library styles (import 'vue-renderer-markdown/index.css').
2. Confirm your bundler includes static asset imports (SVG files) from dependencies.
3. If you override icon components, verify your custom replacements render the expected SVG output.

Performance issues with large documents

Symptoms: Slow rendering, high memory usage, or UI lag with large Markdown files (>10k lines).

Solutions:

1. Use streaming rendering to update content incrementally instead of replacing the entire content at once.

2. Enable pre-rendering for code blocks if you don't need Monaco editor:

   <MarkdownRender :content="markdown" :render-code-blocks-as-pre="true" />

3. Limit Mermaid diagram complexity or consider pre-rendering complex diagrams server-side.

4. Use MarkdownCodeBlockNode instead of CodeBlockNode for lighter syntax highlighting:

   import { MarkdownCodeBlockNode, setCustomComponents } from 'vue-renderer-markdown'
   
   setCustomComponents({
     code_block: MarkdownCodeBlockNode
   })

Math formulas not rendering correctly

Symptom: Math formulas show raw LaTeX or render incorrectly.

Solutions:

1. LaTeX math rendering is optional and requires katex to be installed by the host application. If you need math rendering, install katex and import its stylesheet (see the "Important Notes" above). Ensure you're using valid LaTeX syntax:

• Inline math: $a^2 + b^2 = c^2$
• Block math: $$\int_0^\infty e^{-x^2} dx$$

2. For advanced customization, configure math options:

   import { setDefaultMathOptions } from 'vue-renderer-markdown'
   
   setDefaultMathOptions({
     commands: ['infty', 'perp', 'alpha'],
     escapeExclamation: true
   })

3. See the Math rendering options section for detailed configuration.

Still having issues?

• Check the GitHub Issues to see if someone else has encountered the same problem
• Open a new issue with:
  • Your environment (Node version, framework, bundler)
  • Minimal reproduction code
  • Console errors or screenshots
  • Steps to reproduce the issue

Thanks

This project is built with the help of these awesome libraries:

• stream-monaco — A framework-agnostic library for integrating Monaco Editor with Shiki highlighting, optimized for streaming updates
• stream-markdown — Streaming code/markdown highlighter utilities (Shiki-based)
• mermaid — Diagramming and charting tool that uses Markdown-inspired syntax
• shiki — Syntax highlighter powered by TextMate grammars and VS Code themes

Thanks to the authors and contributors of these projects!

Star History

License

MIT © Simon He

Vite Configuration & Worker Usage (Important!)

If you're using Vite, you only need the following minimal configuration:

import vue from '@vitejs/plugin-vue'
// vite.config.ts
import { defineConfig } from 'vite'

export default defineConfig({
  plugins: [vue()],
  worker: {
    format: 'es',
  },
})

Why? You now manually import and set workers via API, so Vite's optimizeDeps is not needed for worker files. This avoids pre-bundling issues and makes configuration simpler.

How to use workers in your app (Vite example)

You must import the worker using Vite's ?worker syntax and inject it into the library via the API:

// main.ts or your app entry
import KatexWorker from 'vue-renderer-markdown/workers/katexRenderer.worker?worker'
import { setKaTeXWorker } from 'vue-renderer-markdown/workers/katexWorkerClient'

// For Mermaid:
import MermaidWorker from 'vue-renderer-markdown/workers/mermaidParser.worker?worker'
import { setMermaidWorker } from 'vue-renderer-markdown/workers/mermaidWorkerClient'

setKaTeXWorker(new KatexWorker())

setMermaidWorker(new MermaidWorker())

You do not need to pass the worker via component props. Just call the API once in your app entry before using Markdown components.