Blume

The open-source docs framework for humans and agents. Fast, AI-ready, and zero-config. Free and open source, forever.

Drop Markdown or MDX into a folder, start the dev server, and get a production-grade docs site — navigation, search, theming, Open Graph images, and a rich component library — with no app boilerplate to write or maintain. Blume generates and drives a hidden Astro project for you; run npx blume eject to get a standalone Astro app whenever you want full control.

Documentation · Quickstart · Components · CLI

Quickstart

Blume needs Node.js 22.19 or newer and a content folder with at least one .md/.mdx file — there's nothing else to set up.

npx blume init

It scaffolds docs/index.mdx and blume.config.ts, adds dev and build scripts to a new package.json, and installs dependencies. Run the dev server with hot reload:

npm run dev

Build static HTML, with a local search index, into dist/:

npm run build

In a project that already has a package.json, blume init leaves it alone: add "dev": "blume dev" and "build": "blume build" to its scripts, or run npx blume dev. Blume works with any package manager and never requires you to set up Astro or Tailwind yourself.

Moving from another docs framework? npx blume migrate hands a Mintlify, Docusaurus, GitBook, MkDocs, ReadMe, or other docs site to Claude Code or Codex — see Migrate to Blume. On Blume 1, run npx blume@latest upgrade — see Upgrade to Blume 2.

Features

• Zero-config, even the template — a folder of docs is a complete project. No starter to clone, no framework to learn. Configuration is opt-in, one file at a time.
• Fast by default — static HTML on Astro and Vite; the core theme ships no client framework JS so pages score well on Core Web Vitals out of the box.
• Type-safe config — blume.config.ts and every meta.ts are real TypeScript, validated by a schema and authored with defineConfig / defineMeta, so your editor catches mistakes before a build.
• Components, no imports — cards, columns, steps, tabs, accordions, badges, code groups, frames, file trees, type tables, live component previews, diffs, and more, usable in any MDX page.
• Local search — Orama runs in dev and production with no hosted service; FlexSearch, Pagefind, Algolia, Typesense, Orama Cloud, and Mixedbread are one adapter away (search: pagefind() from blume/search).
• AI-ready — llms.txt / llms-full.txt, raw Markdown at any .md URL, a JSON docs API, Copy as Markdown, Open in chat, an optional in-page assistant, and a hosted MCP server so coding agents can search and read your docs directly.
• Agent skills — Blume ships agent skills that teach a coding agent to scaffold, write, and maintain your docs site.
• Content sources — mix local files with remote MDX, GitHub Releases, Notion, Sanity, Contentful, Payload, Strapi, or any custom backend into a single site.
• Internationalization — drop translated files into place for locale-aware routing, per-language navigation, translated UI, and SEO.
• SEO — metadata, Open Graph images (rendered at build with Takumi), sitemap, robots.txt, RSS feeds, and JSON-LD, built in.
• API references — render OpenAPI, AsyncAPI, and GraphQL specs as native reference pages (one per operation, with schemas, auth, and a request playground) with openapi(), asyncapi(), and graphql() from blume/reference, or embed Scalar's UI with scalar().
• Export — let readers download any page as a PDF or EPUB, entirely client-side, so static builds stay static.
• Customization — component overrides, React islands, custom pages, Tailwind v4 theme tokens and theme.css, and a source-component registry (blume add).
• Eject — npx blume eject produces a standalone Astro project that still uses the blume package.

CLI

Run them through your package manager (npx blume ) or a package.json script. See the CLI reference for every flag.

How it works

The Blume CLI loads blume.config.ts, scans your content into a graph, and generates a hidden Astro project under .blume/ that it drives for dev and build. Astro renders through a catch-all page that imports Blume's shipped components, the generated data, and your overrides. .blume/ is regenerated on each run — only changed files are written, so hot reload stays fast — until you blume eject and own it.

Deployment

blume build outputs static HTML to dist/ — deploy to any static host (Vercel, Netlify, Cloudflare Pages, GitHub Pages, S3 + CloudFront, or any CDN). For request-time features like the assistant or the MCP server, name a host adapter from blume/deploy in blume.config.ts, which switches the build to server output:

import { defineConfig } from "blume";
import { vercel } from "blume/deploy";

export default defineConfig({
  deployment: vercel(),
});

On Vercel, Netlify, and Cloudflare Pages the site URL is detected automatically. The adapter never is: name it in blume.config.ts.

Compatibility

Development

This repository is a monorepo: the published package lives in packages/blume, and apps/docs is Blume's own documentation, built with Blume.

bun install
bun run check       # lint + format (Ultracite)
bun run typecheck
bun run test

See CONTRIBUTING.md for architecture and conventions.

License

MIT © Hayden Bleasel