Blume

Product Introduction

1. Definition: Blume is an open-source, static site generator (SSG) and documentation framework built on the Astro and Vite build tooling ecosystem. It is specifically engineered to transform plain markdown and MDX files into high-performance, SEO-optimized, and AI-ready documentation websites.
2. Core Value Proposition: Blume exists to eliminate the friction and complexity traditionally associated with building and maintaining developer documentation. It provides a zero-configuration, markdown-first workflow that automatically generates fast, beautiful, and fully-featured documentation sites with built-in search, theming, internationalization (i18n), and one-command deployment, directly from a folder of content files.

Main Features

1. Markdown-First Content Engine: Blume's core architecture treats markdown as the single source of truth. It automatically infers site structure, routing, and navigation from the file and folder hierarchy within the content/ directory. It supports full Markdown and MDX syntax, including code blocks, callouts, tables, and math equations, rendering them into static HTML at build time for optimal performance. The system uses Astro's content collections for type-safe content management.
2. Zero-Config, Convention-Based Setup: The framework requires minimal initial configuration. Running npx create-blume@latest scaffolds a complete project. Features like sidebar navigation, breadcrumbs, and page outlines are generated automatically from the file structure. Developers can override defaults through a blume.config.ts file, meta.ts files for folder-level configuration, or frontmatter in individual markdown pages, adhering to a "convention over configuration" philosophy.
3. Integrated AI and Machine-Readability Tools: Blume is built for the AI era. It automatically generates an llms.txt file, a standardized plain-text sitemap that makes the entire documentation corpus easily ingestible by Large Language Models (LLMs) and AI coding assistants. It offers an optional, embeddable "Ask AI" chat assistant for pages and can expose a hosted Model Context Protocol (MCP) server, allowing agents like Claude Code or Cursor to directly interact with the docs. The blume eval command provides an AI-powered testing suite for documentation quality.
4. Advanced Static-Site Capabilities: Leveraging Astro, Blume outputs static HTML, CSS, and JavaScript by default, resulting in perfect Core Web Vitals scores. It supports Astro's "islands architecture," allowing developers to place interactive React, Vue, or Svelte components in an islands/ directory for automatic, partial hydration without per-page imports. It also supports server-side rendering (SSR) via Astro adapters for dynamic needs. Features like client-side PDF/EPUB export, OpenAPI spec rendering into native API reference pages, and built-in blog/changelog generators are included.
5. Comprehensive DX and Internationalization: The developer experience is central. It includes a full CLI (blume dev, build, preview), hot module reloading via Vite, and a rich component library (cards, tabs, steps, etc.) usable in MDX. For i18n, Blume supports locale-aware routing, per-language navigation trees, and automated translation workflows via blume translate, which uses AI to sync content across configured locales.

Problems Solved

1. Pain Point: The high overhead and maintenance cost of custom-built documentation sites. Developers and technical writers often spend excessive time configuring build tools (like Next.js, Gatsby), setting up navigation, implementing search (Algolia), and ensuring performance and SEO, rather than writing content.
2. Target Audience: Primary users are Developer Advocates, Technical Writers, and Open-Source Maintainers who need to ship and maintain polished documentation quickly. Secondary users are Product Managers and Startup Founders who need to publish technical product docs without a dedicated web development team. It's also ideal for developers building internal wikis or knowledge bases.
3. Use Cases: Essential for launching the initial documentation for a new API, SDK, or open-source library. Critical for companies scaling their documentation to multiple languages (i18n). Perfect for teams that want to move from a basic README or a cumbersome wiki to a professional, branded, and searchable documentation portal with minimal ongoing DevOps effort.

Unique Advantages

1. Differentiation: Unlike generic static site generators (e.g., Docusaurus, Next.js), Blume is hyper-specialized for documentation, baking in features like navigation generation, component libraries, and AI tooling out of the box. Compared to hosted docs platforms (ReadMe, Mintlify), Blume is open-source, offers greater design control, and avoids vendor lock-in while providing similar ease of use. Its deep Astro integration offers better performance and a more modern DX than older SSGs like Jekyll or Hugo.
2. Key Innovation: Blume's primary innovation is its deep integration of AI-readiness as a first-class feature. The automatic llms.txt generation, built-in eval suite, and MCP server support are not afterthoughts but core architectural decisions. This positions it uniquely for the emerging workflow where documentation must be consumable by both humans and AI agents. Secondly, its "content source" abstraction allows mixing local markdown with remotely fetched content (e.g., from a GitHub repo) seamlessly, enabling novel doc-as-code workflows.

Frequently Asked Questions (FAQ)

1. Is Blume better than Docusaurus for documentation? Blume is often a superior choice for teams prioritizing a modern, Vite-based developer experience, superior performance via Astro's island architecture, and built-in AI-ready features like llms.txt and agent evals. While Docusaurus is React-centric and has a larger plugin ecosystem, Blume offers more out-of-the-box conveniences for documentation-specific needs and greater design flexibility with Tailwind CSS.
2. How does Blume handle search functionality? Blume provides client-side search by default, which works instantly with no API keys or external services, indexing content at build time. As documentation scales, it can be seamlessly switched to hosted, server-side search backends or semantic search providers for better performance and relevance on larger sites, all configurable within blume.config.ts.
3. Can I use Blume for a blog or marketing site, not just docs? Yes, while optimized for documentation, Blume's underlying Astro foundation makes it capable of building any content-focused website. It includes specific guides and features for adding a blog (with RSS feed) and creating fully custom .astro pages, allowing you to mix documentation, blog posts, and marketing landing pages within a single, cohesive project.
4. What is the "blume eval" command used for? The blume eval command implements an AI-powered documentation test suite. It uses a configured AI agent (like Claude or GPT) to answer a set of user questions based solely on the documentation content. A separate "judge" agent then grades the answers. This process can be integrated into CI/CD pipelines to fail builds if the documentation is incomplete or cannot accurately answer critical user queries, ensuring doc quality.
5. Does Blume support deploying to platforms other than Vercel? Absolutely. Since Blume builds standard static assets (or can be adapted for SSR), it can be deployed to any static hosting service, such as Netlify, Cloudflare Pages, GitHub Pages, or AWS S3. The deployment guide provides specific instructions for zero-config deployment on various platforms.