AI/TLDR

HTML Anything

A local HTML editor driven by the coding-agent CLI you already have installed

Generative UI SDKsOpen source
Language
TypeScript
License
Apache-2.0
$git clone https://github.com/nexu-io/html-anything

Overview

HTML Anything is an agent-era HTML editor built on a simple premise: Markdown is the draft, HTML is what people actually read, and in an agentic workflow you no longer hand-edit the document — so the output format should be the one the reader wants. You run it locally, pick a template, describe what you want, and a coding agent writes the page while a sandboxed preview shows the result.

It does not ask for an API key. On startup the app scans your PATH — including the directories a GUI-launched Node process normally misses, such as ~/.local/bin, ~/.bun/bin, /opt/homebrew/bin and ~/.npm-global/bin — and surfaces every coding-agent CLI it recognises, reusing the session you are already signed into. Nine CLIs are detected: Claude Code, Cursor Agent, Codex, Gemini CLI, GitHub Copilot CLI, OpenCode, Qwen Coder, Aider and IBM Bob.

The content side is 75 composable skill templates spread across nine deliverable surfaces: magazine articles, keynote decks, résumés, posters, Xiaohongshu cards, tweet cards, web prototypes, data reports and Hyperframes videos. Each skill is a directory with its own SKILL.md and a real example.html you can open straight from the repo. Finished pages export in one click to WeChat, X or Zhihu, or download as .html or .png. The project comes from the team behind Open Design and is built on top of it.

What it does

  • Zero API key: detects and reuses whichever coding-agent CLI is already signed in on your PATH
  • Nine supported agent CLIs — Claude Code, Cursor Agent, Codex, Gemini CLI, GitHub Copilot CLI, OpenCode, Qwen Coder, Aider and IBM Bob
  • 75 composable skill templates across nine deliverable surfaces, each shipping a runnable example.html
  • Sandboxed preview so generated markup renders without touching the host page
  • One-click export to WeChat, X and Zhihu, or download as .html or .png
  • Local-first: the app, the templates and the agent session all run on your own machine

Getting started

The repo is a small pnpm workspace; the complete Next app lives in next/. Clone it, install, and start the dev server — there is no key to configure, because the app borrows the agent CLI session you already have.

Clone and run the app

Start the Next app from the workspace and open it in a browser.

bashbash
git clone https://github.com/nexu-io/html-anything
cd html-anything
pnpm install
pnpm -F @html-anything/next dev
# → http://localhost:3000

Pick an agent and a template

The top bar auto-detects whichever coding-agent CLI you are already signed in to. Choose a skill template, paste your content, and press ⌘+Enter to let the agent write the page.

Browse the skill catalog

Skills live as directories under the app's template tree, each with a SKILL.md carrying a `recommended:` rank and an example.html you can open without any setup.

texttext
next/src/lib/templates/skills/

Run the workspace checks

Package commands are run from the repo root; the root package.json deliberately does not proxy app or e2e commands.

bashbash
pnpm exec tsx scripts/guard.ts
pnpm -F @html-anything/next typecheck
pnpm -F @html-anything/next test
pnpm -F @html-anything/next build

Commands and code are distilled from the project's own documentation — always check the official repo for the latest.

When to use it

  • Turn a Markdown draft into a magazine-styled HTML article your readers actually get, without hand-writing the layout
  • Generate a keynote deck or poster from a brief using a locked layout system instead of freehand CSS
  • Produce social cards for X, WeChat or Xiaohongshu from one piece of source content and export them as PNGs
  • Use the coding-agent subscription you already pay for as the rendering engine, with no extra API key or billing

How HTML Anything compares

HTML Anything alongside other open-source generative ui sdks tools AI/TLDR tracks, ranked by GitHub stars.

ToolStarsWhat it does
CopilotKit★ 37.4kA frontend framework for adding in-app AI copilots and agent-driven generative UI to React and other apps, and the maker of the AG-UI protocol.
Vercel AI SDK★ 26.9kA TypeScript toolkit for building AI apps and agents that includes streaming helpers and generative-UI features for rendering model output as React components.
OpenUI★ 22.6kOpenUI lets you describe a UI in plain language, see it rendered live, ask for changes, and convert the result to React, Svelte, or Web Components.
AG-UI Protocol★ 16kAn open protocol that standardizes how AI agents stream events and UI to frontend applications, with SDKs across multiple languages and frameworks.
assistant-ui★ 12.2kA TypeScript/React library of composable chat primitives for building AI chat apps that can render tool calls and JSON as interactive React components.
Tambo★ 11.2kA generative UI SDK for React where you register components with Zod schemas and an agent picks the right one and streams its props to the user.
OpenUI (Thesys)★ 9.8kAn open standard and runtime for generative UI that uses a compact streaming language to let models emit interactive charts, forms, tables, and cards.
HTML Anything★ 8.9kA local HTML editor driven by the coding-agent CLI you already have installed