# HTML Anything

> Open-source agentic HTML editor that uses your local coding-agent CLI to turn Markdown, CSV, or JSON into ship-ready HTML for WeChat, X, Zhihu, and Xiaohongshu.

HTML Anything is an open-source agentic HTML editor from the OpenDesign family, published under the nexu-io GitHub organization. It takes Markdown, CSV, TSV, JSON, SQL, Excel, or plain text and uses a coding-agent CLI already signed in on your machine to generate single-file HTML. Output can be exported to WeChat, X, Zhihu, Xiaohongshu, standalone HTML, or PNG.

## What It Is

HTML Anything is a local-first web app (a Next.js shell) that detects coding-agent CLIs on your PATH and reuses their existing sessions, so no separate API key is required. The project's docs list Claude Code, Cursor Agent, OpenAI Codex, Gemini CLI, GitHub Copilot CLI, OpenCode, Qwen Coder, Aider, and IBM Bob. You paste or drop content, pick one of 75 SKILL.md-based templates, and the agent writes the page.

## How the Workflow Runs

The agent's stdout is streamed over server-sent events into a sandboxed iframe, so the layout appears as it is generated and can be interrupted. Tabular input is parsed in the browser with papaparse and xlsx. Each skill is a folder with a SKILL.md, optional assets and references, and an example.html. Adding a skill means adding a folder and restarting the dev server. Skills carry hard constraints such as a CJK-first font stack, an 8 px baseline grid, and contrast of at least 4.5.

## Surface Modes and Export Targets

Nine surface modes are documented: magazine article, keynote deck, résumé, poster, Xiaohongshu card, tweet card, web prototype, data report, and Hyperframes video storyboards. Export options include WeChat MP with CSS inlined via juice, X/Weibo/Xiaohongshu as 2× PNG via the clipboard, Zhihu with LaTeX equations as image placeholders, a standalone .html file, and a high-DPI PNG.

## Current Status

The README describes the project as early but real, with agent detection, the skill picker, streaming render, sandboxed preview, and export marked stable. Multi-template compare and Hyperframes-to-mp4 handoff to Remotion are in progress. A browser extension, history/version diff, and a skill marketplace are planned. The agent runs locally; the web layer can also be deployed to Vercel.

## Features
- Reuses locally authenticated coding-agent CLIs with no extra API key
- Auto-detects supported agent CLIs on PATH
- 75 SKILL.md-based skill templates
- Nine surface modes (magazine, deck, résumé, poster, Xiaohongshu card, tweet card, web prototype, data report, Hyperframes video)
- SSE streaming render into a sandboxed iframe preview
- Input auto-detect for Markdown, CSV, TSV, JSON, SQL, Excel, and plain text
- One-click export to WeChat, X/Weibo, Xiaohongshu, Zhihu, HTML, and PNG
- Hyperframes storyboard frames for Remotion handoff

## Integrations
Claude Code, Cursor Agent, OpenAI Codex, Gemini CLI, GitHub Copilot CLI, OpenCode, Qwen Coder, Aider, IBM Bob, WeChat, X, Zhihu, Xiaohongshu, Remotion, Floatboat

## Platforms
WINDOWS, MACOS, LINUX, ANDROID, IOS, WEB, API, CLI

## Pricing
Open Source

## Links
- Website: https://open-design.ai/html-anything/
- Documentation: https://github.com/nexu-io/html-anything#readme
- Repository: https://github.com/nexu-io/html-anything
- EveryDev.ai: https://www.everydev.ai/tools/html-anything