rehype

rehype is a unified processor preconfigured with rehype-parse and rehype-stringify, giving you a ready-to-use pipeline for parsing HTML into a HAST syntax tree, running plugins to transform it, and serializing it back to HTML. It is a shortcut for unified().use(rehypeParse).use(rehypeStringify), part of the broader unifiedjs collective alongside remark (Markdown) and retext (prose).

As a monorepo, it also houses rehype-parse, rehype-stringify, and rehype-cli, and anchors a large third-party plugin ecosystem used for HTML linting, sanitization, minification, and static-site content transformation.

What You Get

• A preconfigured unified processor combining rehype-parse and rehype-stringify
• A HAST (HTML AST) syntax tree representation for programmatic HTML manipulation
• Compatibility with the large ecosystem of rehype plugins (sanitize, minify, highlight, etc.)
• TypeScript types for the processor and its syntax tree
• A companion rehype-cli package for command-line HTML linting/formatting
• Interop with remark (Markdown) via remark-rehype for Markdown-to-HTML pipelines

Common Use Cases

• Transforming HTML content in static site generators (adding IDs to headings, autolinking, etc.)
• Sanitizing untrusted HTML before rendering it in a web app
• Linting and auto-formatting HTML files from the command line via rehype-cli
• Building a Markdown-to-HTML pipeline together with remark and remark-rehype
• Extracting or rewriting structured data from HTML documents programmatically

Under The Hood

Architecture - The root packages/rehype/index.js composes three sibling packages in the same monorepo: rehype-parse (HTML string to HAST tree), the unified engine (which runs registered plugins over the tree), and rehype-stringify (HAST tree back to an HTML string); rehype-cli sits alongside as a separate consumer of the same processor for command-line use, and the whole ecosystem interops with remark/retext via shared unist-family syntax trees.

Tech Stack - Pure ESM JavaScript ("type": "module") with hand-written .d.ts type declarations, managed as an npm/yarn workspaces monorepo (packages/rehype-parse, packages/rehype-stringify, packages/rehype, packages/rehype-cli), using c8 for coverage, remark-preset-wooorm for internal doc linting, and type-coverage to enforce TypeScript type completeness across the packages.

Code Quality - The monorepo has an extensive test/ directory with parse-error fixtures and syntax-tree assertions via hast-util-assert, plus CI workflows under .github/workflows; the actual rehype package itself is intentionally tiny (a ~10-line index.js), so most complexity and testing responsibility lives in the rehype-parse/rehype-stringify/unified dependencies it composes.

API Design - The processor exposes exactly the unified API (.use(), .process(), .processSync(), .parse(), .stringify()), so anyone familiar with unified/remark can use rehype immediately with zero new concepts to learn; the README explicitly steers users toward using unified directly when they don’t need both parse and stringify, favoring composability over a bespoke convenience API.