react-markdown

Safely render Markdown as React elements, no dangerouslySetInnerHTML required

Library
npm
v10.1.0
15,892 stars
MIT License

Repository Health

Pre-computed score based on development activity, maintenance, community, maturity, and trend momentum. How we score it →
66 /100 Good
Development Activity 60
Maintenance 36
Community 68
Maturity 60
Momentum 40

Technical Analysis

AI-assessed by reading the actual repository — architecture, code quality, innovation, and documentation. How we score it →
85 /100 Excellent
Architecture 82
Code Quality 85
Innovation 88
Learning Curve 85

react-markdown is a React component that turns a Markdown string into a tree of React elements instead of raw HTML, so nothing ever passes through dangerouslySetInnerHTML and the app stays safe from injected scripts by default. Under the hood it delegates parsing and transformation to the unified ecosystem (remark for Markdown, rehype for HTML), which means it inherits a mature plugin system, full CommonMark compliance, and optional GitHub Flavored Markdown support via remark-gfm.

Beyond safety, its main draw is customizability: any HTML tag produced by the render can be swapped for a React component via the components prop, letting consumers replace <a>, <img>, or <code> with app-specific implementations (routed links, lazy images, syntax-highlighted code blocks) without forking the renderer. Three entry points cover different rendering needs — a synchronous Markdown component for the common case, MarkdownAsync for server-side async plugins, and MarkdownHooks for client-side async plugins via React hooks.

What You Get

  • A drop-in <Markdown> component that renders a Markdown string prop straight to React elements
  • Automatic protection against script injection — HTML in the source markdown is stripped or escaped by default via a safe unified pipeline
  • A components prop to remap any rendered tag (headings, links, images, code blocks, tables) to your own React components
  • remarkPlugins and rehypePlugins arrays for hooking into the wider unified plugin ecosystem (GFM tables/strikethrough, footnotes, math, syntax highlighting, table of contents, and more)
  • MarkdownAsync and MarkdownHooks variants for rendering with async plugins on the server or client respectively
  • A configurable urlTransform hook (with a safe default) to sanitize or rewrite every URL-bearing attribute in the output

Common Use Cases

  • Rendering user-authored Markdown (comments, README previews, CMS content) safely in a React app without hand-rolling XSS sanitization
  • Building a docs or blog site where Markdown needs to render through the app’s own design-system components instead of raw HTML tags
  • Adding GitHub Flavored Markdown features (tables, task lists, strikethrough, autolinks) via remark-gfm on top of the base renderer
  • Rendering LLM chat responses that come back as Markdown, with custom components for code blocks, links, and citations

Under The Hood

Architecture — The package exports Markdown (sync), MarkdownAsync (async, server-safe), and MarkdownHooks (async via React hooks) from lib/index.js. Internally it builds a unified() processor chaining remarkParse → user remarkPlugins → remarkRehype (with allowDangerousHtml: true so raw HTML survives as a raw node instead of being silently dropped) → user rehypePlugins. createFile() wraps the Markdown string in a VFile. After processing, a post() step runs a single unist-util-visit pass over the resulting hast tree to: reject deprecated pre-v6 props with actionable errors (via devlop’s unreachable()), convert or strip raw HTML nodes depending on skipHtml, rewrite every URL-bearing attribute through urlTransform/defaultUrlTransform (a protocol-allowlist regex), and filter elements via allowedElements/disallowedElements/allowElement. Finally hast-util-to-jsx-runtime’s toJsxRuntime() turns the filtered hast tree into React elements using the automatic JSX runtime, passing the original hast node to matched custom components via passNode. MarkdownHooks wraps this same pipeline in useMemo/useEffect/useState with a cancellation guard, deferring async-plugin work to the client without blocking initial render.

Tech Stack — A single index.js re-export plus a ~450-line lib/index.js, authored as pure ESM ("type": "module") checked-JS-with-JSDoc rather than .ts source, with tsc emitting the shipped .d.ts. Runtime dependencies are entirely from the unified/remark/rehype ecosystem (unified, remark-parse, remark-rehype, hast-util-to-jsx-runtime, html-url-attributes, unist-util-visit, vfile, devlop), plus type-only @types/hast/@types/mdast. React is a peer dependency only (>=18), never bundled. Dev tooling is Node’s built-in test runner (not Jest/Vitest), xo/prettier for lint and format, c8 for coverage, and type-coverage for strict type enforcement.

Code Quality — test.jsx is a single 1,280-line, 87-case suite covering the three exported components, deprecated-prop error paths, element allow/deny-listing, custom components, plugin composition, URL transforms, and SSR streaming, run via node:test with global-jsdom and @testing-library/react. The test script gates on c8 --100 (100% statement coverage) and type-coverage --at-least 100 --strict. Naming is terse and consistent (createProcessor/createFile/post), and error handling favors explicit, descriptive unreachable() throws that point at a changelog anchor for every removed/renamed legacy option rather than failing silently.

API Design — The public surface is deliberately small: a default Markdown export plus two async variants, so the common case is <Markdown>{md}</Markdown> with zero required configuration. Extensibility layers on cleanly via remarkPlugins/rehypePlugins arrays (the same plugin objects used across the broader unified ecosystem) and a components map that mirrors JSX.IntrinsicElements, so anyone familiar with remark/rehype faces almost no new API surface. Deprecated pre-v6 props throw actionable errors instead of silently misbehaving — unusually good DX for a library with a long breaking-change history. The readme is extensive, with runnable examples for plugins, custom components, and math/syntax-highlighting integrations.

Used by 164 apps in this directory

Python
58%
Apache 2.0

MLflow

AI Development · Monitoring

28,154

The open source AI engineering platform for debugging, evaluating, monitoring, and optimizing production LLMs and agents at scale.

View details
97
Repo Health
86
Technical
66
Dependency
Built with
Python 58%
TypeScript 33%
Updated 1 weeks ago
Python
66%
Other

Morphik

AI Development · Databases · Search

3,716

Morphik is an AI-native ingestion and retrieval engine that lets developers store, search, and reason over visually rich documents — scanned PDFs, manuals, slides, and video — without duct-taping together OCR, an embedding model, and a vector database.

View details
60
Repo Health
71
Technical
67
Dependency
Built with
Python 66%
TypeScript 23%
Updated 2 weeks ago
Go
56%
Other

Multica

AI Assistants · AI Development

51,481

Turn coding agents into real teammates — assign issues, track progress, and compound reusable skills across a vendor-neutral, self-hosted platform.

View details
87
Repo Health
83
Technical
67
Dependency
Built with
Go 56%
TypeScript 38%
Updated 1 weeks ago
TypeScript
96%
Apache 2.0

Next AI Draw.io

AI Design Tools · Design Tools · Developer Tools

36,063

Turn natural language into professional draw.io diagrams with AI, cloud icons, and an MCP server for your IDE.

View details
78
Repo Health
80
Technical
68
Dependency
Built with
TypeScript 96%
Updated 1 weeks ago
TypeScript
49%
MIT

Nginx Proxy Manager

Developer Tools · Networking · Security

34,253

Manage Nginx reverse proxies and free Let's Encrypt SSL through a beautiful web interface — no Nginx expertise required.

View details
92
Repo Health
77
Technical
73
Dependency
Built with
TypeScript 49%
JavaScript 46%
Updated 1 weeks ago
TypeScript
99%
Other

NocoBase

Low Code Platforms · No Code Platforms

24,377

Open-source AI + no-code platform that lets coding agents and people collaborate to build business systems fast on proven infrastructure.

View details
94
Repo Health
81
Technical
63
Dependency
Built with
TypeScript 99%
Updated 1 weeks ago
Go
93%
MIT

NornicDB

AI Development · AI Memory · Databases

887

A single graph+vector+temporal database for AI workloads — Neo4j-compatible, sub-millisecond hybrid search, and built-in memory decay.

View details
80
Repo Health
82
Technical
71
Dependency
Built with
Go 93%
Updated 1 weeks ago
Go
69%
MIT

Ollama

AI Development · Developer Tools

181,830

Run Llama, Gemma, DeepSeek, and other open LLMs on your own machine with one command and an OpenAI-compatible API.

View details
90
Repo Health
85
Technical
67
Dependency
Built with
Go 69%
C 22%
Updated 1 weeks ago
TypeScript
95%
MIT

OmniRoute

AI Agents

70,824

A free, open-source AI gateway connecting Claude Code, Codex, Cursor, Cline, and Copilot to 237 AI providers (90+ free) through one endpoint, with automatic fallback and token-compression to stretch free-tier limits further.

View details
86
Repo Health
70
Technical
71
Dependency
Built with
TypeScript 95%
Updated 1 weeks ago

Join founders buildingwith open source

Opinionated takes, migration guides, cost-saving tips, and insights from the open source ecosystem.

Subscribe on Substack
Join 750+ subscribers