markstream-vue

Why we included this project

Anyone who has built a chat UI on top of an LLM token stream knows the raw output looks wrong until the response finishes: partial Markdown flickers and diagrams look broken, while long answers keep shifting the page as they grow. Markstream fixes that in-between state by rendering incomplete Markdown as a stable, progressively updating document, so a half-written sentence stays readable and a growing Mermaid diagram fills in rather than popping into place. The renderer ships as separate packages for Vue, React, Svelte, Angular, and even Vue 2, so teams can keep the same streaming behavior when they move between frameworks. It also handles the heavier parts of AI output: streaming code with Shiki or Monaco, KaTeX math, safe HTML instead of raw v-html, and low-jitter updates for long responses. If your product consumes SSE or WebSocket streams and you want the rendered chat to keep pace with the tokens, this is worth testing before you hand-roll your own Markdown handling.