vsmarked.

The Engine Under Our Hood

get started ↗ examples all comparisons

FIG-001

SHEET 01 / 07

FIG-002 / OVERVIEW

at a glance.

Compare marked and @humanspeak/svelte-markdown: direct markdown parsing versus a Svelte 5 component layer with renderers, caching, and safety controls.

FIG-003 / FEATURE MATRIX

side-by-side.

Every surface that matters, compared without spin.

feature	@humanspeak/svelte-markdown	marked
Svelte 5 Native	yes	no
TypeScript Support	yes	yes
Component Output	Svelte components	Raw HTML string
Custom Renderers	Svelte components	Token manipulation
Token Caching	Built-in LRU cache	Manual implementation
HTML Safety	Allow/deny per tag	Manual sanitization
LLM Streaming Mode marked is the parsing engine only; incremental streaming behavior must be built manually on top.	writeChunk() / resetStream()	no
Streaming HTML Output marked parses to tokens or HTML strings per call; streaming reconciliation of nested HTML across chunks is left to the consumer.	Partial blocks reconcile when </tag> arrives	no
Extensions	Full marked extensions support	Full extensions API
Math (KaTeX)	Built-in extension (markedKatex)	Via marked-katex-extension
Diagrams (Mermaid)	Built-in extension (markedMermaid)	Via marked-mermaid
GitHub Alerts	Built-in extension (markedAlert)	Via marked-alert
Footnotes	Built-in extension (markedFootnote)	Via marked-footnote
GFM Support	yes	yes
Bundle Size	~15KB (includes marked)	~8KB (parser only)
Framework Integration	Native Svelte	Requires {@html}

FIG-004 / STRENGTHS

where each shines.

▣ svelte markdown 12

+Svelte 5 runes-native — built for runes, not retrofitted
+TypeScript-first with full type safety
+Built-in token caching (50-200x faster re-renders on repeated content)
+LLM streaming with imperative writeChunk() / resetStream() API
+23 markdown renderers + 84 HTML tag renderers — every override is a Svelte snippet
+First-class extensions: KaTeX math, Mermaid diagrams, GitHub alerts, footnotes
+Built-in XSS protection — protocol allowlist, event-handler stripping, attribute sanitization
+Allow/deny utilities (allowHtmlOnly, excludeRenderersOnly, etc.) for fine-grained control
+Drop-in component — works anywhere in your Svelte app
+All the power of marked, wrapped in Svelte components
+Full marked extensions API pass-through
+No {@html} needed — renders as safe Svelte components

▢ marked 5

+Most popular JS markdown parser (38M+ weekly npm downloads)
+Smaller bundle (parser only, no rendering layer)
+Framework-agnostic — use anywhere
+Maximum flexibility with token/renderer hooks
+Extensive extension ecosystem

FIG-005 / LIMITATIONS

where each falls short.

▣ svelte markdown 3

−Smaller community (newer project)
−Runtime-only (no build-time optimization)
−Slightly larger bundle (includes rendering layer on top of marked)

▢ marked 4

−Outputs raw HTML — requires {@html} and manual XSS handling
−No component-based rendering in Svelte
−Must manually implement caching
−No allow/deny controls for HTML tags

FIG-006 / VERDICT

the honest call.

We use marked internally — so you get all of its parsing power for free. Choose marked directly when you need a framework-agnostic parser or want the absolute smallest bundle. Choose @humanspeak/svelte-markdown when you want the Svelte DX layer: component rendering, caching, HTML controls, and TypeScript types.

FIG-007 / MORE