contextmd-cli

🧠 ContextMD

Feed your Agents the Context they deserve.

🚀 maintain your AI Documentation effortlessly?

Check out SuperDocs.cloud — The platform to build, host, and maintain AI-ready documentation in 1-click directly from your GitHub repository.

- Always up to date. Zero config. Production ready.

ContextMD is the ultimate terminal utility for turning complex documentation websites into a single, high-density AI Context File.

Modern LLMs and Agents (like Claude 3.5 Sonnet, GPT-4o, or Gemini 1.5 Pro) are powerful, but they struggle to navigate multi-page documentation sites effectively. They get lost in navigation bars, footers, duplicate content, and fragmented pages.

ContextMD solves this. It crawls, cleans, and chemically refines entire documentation sites into a single context.md file that you can drop directly into your LLM's context window.

☁️ Go Pro with SuperDocs.cloud

Love the CLI but want automation?

SuperDocs.cloud takes this concept to the enterprise level:

• 🔄 Auto-Sync: Automatically updates your context whenever you push to GitHub.
• 🌍 Hosted URLs: Get a permanent, shareable URL for your documentation context.
• 🧠 Smart Versioning: maintain multiple versions of your docs (v1, v2) context.
• ⚡ 1-Click Setup: Just connect your repo, and we handle the scraping, cleaning, and hosting.

Start Building on SuperDocs.cloud →

✨ Features

• 🕷️ Deep Crawling: Intelligently traverses documentation sites, following links and building a comprehensive map of the content.
• 🧠 AI-Powered Refinement: Uses OpenAI's models (configurable) to "read" each page and rewrite it for machine comprehension, stripping fluff and prioritizing logic, API signatures, and examples.
• 🧹 Noise Reduction: Automatically detects and separates main content from sidebars, headers, footers, and advertisements.
• ⚡ High Performance: Concurrent processing with a beautiful, real-time CLI dashboard.
• 📄 Single File Output: Produces a consolidated Markdown file with clear headers and structure, perfect for RAG systems or direct LLM context.

🚀 Installation

Ensure you have Node.js 18+ installed.

Global Install (Recommended)

npm install -g contextmd-cli

Run via npx (No install required)

npx contextmd https://docs.example.com

🛠️ Usage

Quick Start

• Get an OpenAI API Key: ContextMD uses AI to compress and refine the content.
• Run the tool:

export OPENAI_API_KEY=sk-proj-...
contextmd https://docs.turso.tech

This will generate a context.md file in your current directory.

Command Line Options

Usage: contextmd [options] <url>

Arguments:
  url                      Base URL of the documentation to convert

Options:
  -k, --key <key>          OpenAI API Key (can also be set via OPENAI_API_KEY env var)
  -o, --output <path>      Output file path (default: "context.md")
  -l, --limit <number>     Max pages to crawl (default: "100")
  -h, --help               display help for command

Examples

Crawl a specific documentation site with a page limit:

contextmd https://developer.spotify.com/documentation/web-api --limit 50

Save to a specific location:

contextmd https://stripe.com/docs/api -o ./stripe-context.md

🏗️ How It Works

ContextMD operates in a three-stage pipeline:

• The Crawler:

  • Starts at the provided url.
  • Uses a Breadth-First Search (BFS) algorithm to find internal links.
  • Filters out external links, social media, and irrelevant pages.
  • Respects the --limit flag to prevent infinite loops on massive sites.

• The Processor (The "Brain"):

  • Downloads the raw HTML of each discovered page.
  • Uses turndown and cheerio to convert HTML to Markdown.
  • AI Step: Sends the raw Markdown to an LLM with a specialized system prompt designed to:
    • Summarize verbose sections.
    • Preserve code blocks and API schemas exactly.
    • Remove marketing fluff.
    • Standardize formatting.

• The Compiler:

  • Stitches all processed pages into a single context.md file.
  • Adds a metadata header and table of contents structure (implicitly via markdown headers).

📦 For Developers

Want to build this from source?

• Clone the repo:

  git clone https://github.com/UditAkhourii/contextmd.git
  cd contextmd
  

• Install dependencies:

  npm install
  

• Build:

  npm run build
  

• Run locally:

  node dist/index.js https://example.com
  

🤝 Contributing

We welcome contributions! Please open an issue or submit a PR if you have ideas for:

• Support for local LLMs (Ollama, etc.)
• Better crawling heuristics for SPA (Single Page Apps).
• Output formats (JSON, JSONL for fine-tuning).

📄 License

CC BY-NC 4.0 © Udit Akhouri